Studies of economic voting are plentiful and have established that economic evaluations influence electoral outcomes all over the world. Recently however, a few studies have appeared that question the conventional models of economic voting by asserting that economic perceptions are endogenous and in fact instead largely influenced by party sympathy and vote choice. This challenges the very core of traditional models of economic voting. This study examines the degree of partisan bias in economic perceptions in Sweden and how different models affect affect conclusions. By using several different high quality data sources systematically collected over a long period of time the study can asses both the accuracy of economic perceptions in relation to real economic changes and to what extent partisanship distorts economic perceptions, as well as how the choice of research design affects the results. Two different data sources are used: Firstly, 25 years of consecutive annual cross-sectional surveys are analysed, and secondly, panel data from 8 consecutive parliamentary elections. The results indicate that all data sources show some degree of partisan bias, but that it is often relatively minor in importance compared to the influence of real economic changes. Further, the conclusions concerning the importance of the economy for vote choice differ depending on whether cross-sectional data or panel data is used.
