Yet Another Model of Vote Choice Under Uncertainty
∗
Joan Serra
University of Chicago
Department of Political Science
Abstract
Current models of elections assume that voters perfectly know their ideal points in the
electoral dimensions. This assumption, however, violates the empirical evidence. In this
paper I outline a model that avoids this assumption but includes the actual uncertainty of
voters about their own ideal points. Unlike existing models the one proposed here is able
to simultaneously account for voters’ preference for candidates’ precision or ambiguity,
their candidate choice, and their observed patterns of voting or abstaining.
1
Introduction
Current electoral models assume voters perfectly know their ideal points in the electoral di-
mensions. Even probabilistic models, that include uncertainty in the voter’s calculation, just
allow for voters to be uncertain about the position of the candidates on the electoral space.
The available empirical data, however, does not support this assumption. In the only explicit
existing evidence, Alvarez (1998: 66: Table 4.2) reports that, if asked, a significant propor-
tion of voters admit to be uncertain about their own preferred position on the Liberalism-
Conservatism scale. Only 35.6 percent of the sample he reports say they are certain of their
own placement, while the others say they are somewhat certain (29.9), not certain (6.8), or
even were unable to place themselves on the scale (27.7). This level of certainty about their
∗
Prepared for delivery at the 2005 APSA Annual Meeting, September 1-4, Washington,
DC. Comments welcome at ## email not listed ##.
1
Page 1 of 11 
Convention