Reasons for vote switching: Do new instruments improve our understanding of the floating vote?

Attempts to analyze the reasons for vote switching between consecutive elec tions have been in general quite disappointing. On the one hand, this resul t may be attributed to idiosyncrasies of vote switching; on the other hand, it may be a product of not fully adequate research strategies. This paper explores the second possibility. First, we suggest to study switchers not a s a whole, but to distinguish different groups of switchers according to th e political relevance of the different kinds of switching. Secondly, we dis cuss whether the analysis of switching could benefit from the use of panel data instead of cross-sectional data: the operationalization of switching b y panel data should be superior to the widely used recall method. The hypot heses derived in the theoretical discussion receive strong support from emp irical evidence from the German general election of 1998: both the differen tiated measurement of switchers and the use of panel data enhance analysis of switching based on cleavage theory and the Michigan model. More generall y, we conclude that our understanding of this dynamic element in electoral politics depends on the methodological adequacy of research.