Expectations about voter choices

The only essential differences in the examples analyzed above are in the assumptions that are used to derive or directly specify the choice behavior that the candidates expect from any given voter, conditional on his utility function. When the central behavioral assumption from the traditional models was used, the equilibrium was a median outcome—a conclusion directly analogous to what happens in the corresponding (unidimensional) traditional voting models. When the central behavioral assumptions from the probabilistic voting model developed in Coughlin and Nitzan [1981] were used, the equilibrium was at the maximum of the expected log-utility function—a conclusion directly analogous to what happens in the model in Coughlin and Nitzan [1981]. Indeed, it was only when the corresponding assumptions from Ledyard [1984] were used (i.e., the particular assumptions that Ledyard used to derive the choice behavior that the candidates expect from any given voter, conditional on his utility function) that the conclusions were substantially different from the conclusions derived from earlier models. This reflects the important role these assumptions play in making Ledyard's conclusions about candidate strategies substantially different from earlier results in the literature on spatial models of elections. Copyright Martinus Nijhoff Publishers 1984