Party Ambiguity and Voter Uncertainty. A Re-Assessment

The spatial theory of voting has frequently been criticized for assuming too much of the average voter: voters need to identify their personal ideological or policy preferences and they need to be able to assess the positions of competing candidates or parties. Grofman's discounting models brings the political "status quo" back in and thus requires even more: voters are supposed to factor in how far candidates or parties, if elected, wil be able to move the status quo towards their advertised spatial positions. Theoretically, the discounting model is a very compelling approach. It is able to explain patterns of polarizing trusteeship when party elites factor in discounting and assume much more "extreme" policy stances than their electoral constituencies. The discounting model may also provide explanations for protest voting, electoral, and political polarization from within the spatial modeling tradition. Empirical applications, however, have been rare and often based upon studies of individial cases or elections. This contributions aims at closing this critical research gap. We focus on rich evidence from the first four waves of the CSES project and apply Bayesian statistical tools so as to probe the added value of discounting party positions.