The European Voter Project

by Jacques Thomassen

The purpose of the European Voter project is to systematically describe and explain the electoral changes that have occurred in a number of West-European countries in the second half of the twentieth century. As such it intends to fill an obvious gap in European comparative electoral research. 

At first sight electoral research is one of the best-developed and best-integrated sub-disciplines in political science. A growing number of countries have an established program of academically directed election studies, based on national probability samples of the electorate. Between the end of the 1950s and the early 1970s a national program of election studies was established in the Scandinavian countries, Britain, Germany and the Netherlands. Other European countries followed later.

All these national election studies had two characteristics in common.

  • First, from the very beginning they were set up as a time series, i.e. a core of survey questions is replicated at each election, and
  • Secondly, all these election studies were based on more or less the same theoretical framework and methodology, partly based on a sociological approach of political behaviour as developed by Rokkan a.o. and partly derived from the so-called ‘Michigan school’ of electoral research.

Mainly because of these common intellectual roots the similarity of most of these studies, both in their theoretical and methodological features is striking. This similarity suggests a well-developed program of comparative research. However, such a program does hardly exist. Despite the common intellectual roots of most European election studies, really comparative research is remarkably rare. Partly this is due to the availability of data. For a long time a number of logistical problems bedevilled using existing data in international comparative electoral research. To mention only the most obvious one: for a number of studies written documentation, even the questionnaires, were not available in English.

Awareness of these problems led to the initiative to establish the International Committee for Research into Elections and Representative Democracy (ICORE). It was founded at the end of the 1980s, its founding fathers being the study directors of some of the older national election studies programs in Western Europe, in particular Norway, Sweden, Britain, the Netherlands, Denmark, and Germany. The aim of ICORE is to promote cross-national research into electoral behaviour and representative democracy. One of ICORE’s first priorities was to overcome the logistical problems making the use of existing data difficult¹. In order to solve them a European Elections Database was created, consisting of all the national election studies documented in English. As a result of these efforts the data of most of the major European election studies, documented in English, are now available at the Central Archive in Cologne². In order to facilitate the systematic study of trends in electoral behaviour and its correlates across countries, comparable measures across time within countries for all key concepts were developed. 

This resulted in The European Voter Data Base. The GESIS Data Archive for the Social Sciences (formerly: Central Archive for Empirical Social Research (ZA) at the University of Cologne) is the archive for this data base (GESIS Study Number ZA3911). The data and additional information has been made available by the principal investigators and their teams. All data and documentation relevant for “The European Voter” have been integrated and stored in a data base under the ZA-Codebook Explorer by ZA staff. The national subsets and integrated subsets of the national election studies are available for interested researchers at the same level as the principal investigators. This data base laid the foundation for the book project The European Voter³. The conceptual framework and the organising principle of this book are the following: 

In line with the theory of modernisation the main proposition tested in the book is that over time the explanatory power of the variables in this causal scheme gradually shifts from left to right. Due to changes both in the composition of the electorate and the relationship between social position and electoral behaviour, the once strong relationship between social structure and politics, between social position and party choice will diminish. For the same reason it will no longer be evident that people for their life time will be loyal to a particular party. Ideologies reflecting traditional cleavages will become less important as a factor determining people’s political attitudes and party choice. Voters will decide from election to election what party they will vote for, taking into account the issues of the day, the performance of the incumbent government and their confidence in individual political leaders.

However, this proposition is systematically confronted with an alternative explanation. It has been argued that modernisation theory is wrong in assuming that changes in electoral behaviour are due to an autonomous societal process, independently from the political-institutional context. Therefore, the alternative explanation tested throughout the book is that differences in the development of electoral change between countries and fluctuations within countries are entirely, or at least partly, due to differences or changes in the political-institutional context.

Mochmann, Ekkehard; Oedegaard, Ingvill C.; Mauer, Reiner, ICORE Inventory of National Election Studies: Belgium, Denmark, France, Germany, Hungary, Netherlands, Norway, Sweden and United Kingdom. Bergisch Gladbach: Edwin Ferger Verlag 1998.

Jacques Thomassen (ed.), The European Voter, Oxford: Oxford University Press, 2005 (

  • Denmark
  • Johannes Andersen, Aalborg University
    Jörgen Goul Andersen , Aalborg University
  • Germany
  • Hans-Dieter Klingemann, Wissenschaftszentrum Berlin
    Hermann Schmitt, Mannheimer Zentrum für Europäische Sozialforschung,
    Bernhard Wessels, Wissenschaftszentrum Berlin, Germany
    Tanja Binder, Wissenschaftszentrum Berlin
  • Great Britian
  • John Curtice, University of Strathclyde
  • The Netherlands
  • JacquesThomassen, University of Twente
    Kees Aarts, University of Twente
    Cees van der Eijk, University of Amsterdam
    Pieter van Wijnen, University of Twente
  • Norway
  • Bernt Aardal, Institute for Social Research
    Oddbjörn Knutsen, University of Oslo
    Ola Listhaug, Ola, Norwegian University of Science and Technology Trondheim
    Frode Berglund, Institute for Social Research
  • Sweden
  • Sören Holmberg, University of Göteborg
    Henrik Oscarsson, University of Göteborg
    Maria Oskarson, University of Göteborg
    Staffan Kumlin, University of Göteborg
  • Sex, age, civil status, urbanisation, education, region, religion, church attendance, occupation, working, sector, subjective class, occupational group, income, trade union member, own house/appartment. Political interest, political discussions, efficacy, party attachment, party sympathy, party leader sympathy, party left-right placement, economic evaluation, important issues. Value dimensions: state, moral, authortiarian/libertarian, growth/ecology, Inglehart-Index, when decide to vote, voting behaviour.
  • Denmark:22291 C / 137 V
  • Germany:19018 C / 137 V
    Great Britain:30132 C / 143 V (separate cross-section datasets)
    The Netherlands:19253 C / 96 V
    Norway:16393 C / 130 V
    Sweden:34496 C / 306 V
    Format:SPSS Files
  • Denmark: The weight variables has been included where possible, however, no continuity exists between the weight variable for the separate election studies. In the case of the 1998 election study two separate variables has been coded called weigde98 and weigpo98, the former weighting to achieve demographic representativity, the latter for party representativity.
  • Germany:The weighting factors (weight1) are used to improve demographic representativeness (mostly based on the joint distribution of the criteria of age and sex in the universe and the sample). Weight in panel studies seem to be calculated for the overall “sample” rather than for respondents of individual waves of surveys. A weight to improve only the east west-proportionality is given in 1994 and 1998 additionally.
  • Weight 3 is a political weight which corrects the relationship between non-voter and voters of particular parties according to the official election results.
  • Great Britian: The Netherlands:See documentation in study description on Dutch continuity file Norway: Advised not to apply weights Sweden: Advised not to apply weights