Structure and availability of databases

Methodology

Special attention is devoted in the electoral surveys to methodology. Several focus areas can be identified here :

• Explaining internal vote shifts between two elections (based on loglinear analyses of transition tables).
• Developing causal models which attempt to explain voting behaviour on the basis of political attitudes and socio-demographic characteristics.
• Improving the quality of scale constructions and question wordings (using confirmatory factory analysis and latent class analysis).
• Longitudinal analyses of the panel data gathered during the first two surveys (using specific techniques for longitudinal data analyses)
• Comparative research into a range of causal analysis techniques (comparison between loglinear path analysis and linear covariance models for ordinal variables).
• Studying non-response in surveys.
• Measuring method effects, e.g. measuring 'acquiescence'.
• Research into the influence of the media and time on reporting of the reason for the voting behaviour (based on the 1995 exit poll).
• Research into the cross-cultural equivalence of measurement models.
• Modelling open questions in electoral surveys (through the use of HICLAS analysis).
• Response stability in quasi-balanced scales on political alienation.

Substantive research

For several decades, social scientists in a number of Belgium's neighbouring countries have been able to draw on data from regularly recurring electoral surveys. These surveys ask a representative sample of the population about their political and social ideas and attitudes. Data of this sort form an important source of information both for policymakers and the social scientists themselves. Until the early nineties there was virtually no tradition in Belgium of electoral research at national level based on specific surveys. As a result, few authoritative data were available on voting behaviour at Belgian elections.

In order to bridge this gap, it was decided on the initiative of the Department of Scientific Policy Planning (LPWB) to set up the "Interuniversity Centre of Political Opinion (ISPO)/Point d'Appui Interuniversitaire sur l'Opinion publique et la Politique (PIOP)". ISPO is based at the Sociology Department of the Catholic University of Leuven (K.U. Leuven), while PIOP is housed at the "Departement des Sciences Politiques et Sociales at the Université Catholique de Louvain (U.C. Louvain). In order to safeguard the interuniversity character of the research project, a steering committee was installed comprising various members of Flemish and Walloon universities who are concerned with political or sociological research.

In addition to the above assignment, both ISPO and PIOP develop research projects independently, each in its own domain. The regional centres were given the task of conducting large-scale surveys after each general election to ascertain the attitudes and behaviour of the Belgian electorate. These surveys would then form the basis for building up the databases further for longitudinal analyses, with a view to identifying short-term and long-term changes in the political and social attitudes of the population. An important focus of attention within the two centres is the gathering of high-quality data. This means that there is a strong emphasis on thorough training of interviewers, the careful building up of a regional sampling plan and continual evaluation and adjustment of both the questionnaire as a whole and the specific measurements of the basic concepts within the survey. The data gathered in this way are then made available to the scientific community for further research relating to Belgian elections in both a national and an international context.

download
'Disproportionaliteit van het Belgische Kiessysteem'
van Patrick Vander Weyden
'D'Hondt and Alternative D'Hondt for Two-Tier Districting Systems.
The Belgian Electoral System'.
Order IPSoM-Bulletin 2001/5 contact Patrick Vander Weyden