Information heterogeneity and voter uncertainty in spatial voting: the U.S. presidential elections, 1992-2004
This dissertation addresses voters' information heterogeneity and its effect on spatial voting. While most spatial voting models simply assume that voter uncertainty about candidate preferences is homogeneous across voters despite Downs' early use of uncertainty scale to classify the electorate, information studies have discovered that well and poorly informed citizens have sizeable and consistent differences in issue conceptualization, perception, political opinion and behavior. Built upon the spatial theory's early insights on uncertainty and the findings of information literature, this dissertation claims that information effects should be incorporated into the spatial voting model. By this incorporation, I seek to unify the different scholarly traditions of the spatial theory of voting and the study of political information. I hypothesize that uncertainty is not homogeneous, but varies with the level of information, which are approximated by political activism as well as information on candidate policy positions. To test this hypothesis, I employ heteroskedastic probit models that specify heterogeneity of voter uncertainty in probabilistic models of spatial voting. The models are applied to the U.S. presidential elections in 1992-2004. The empirical results of the analysis strongly support the expectation. They reveal that voter uncertainty is heterogeneous as a result of uneven distributions of information and political activism even when various voting cues are available. This dissertation also discovers that this heterogeneity in voter uncertainty has a significant effect on electoral outcomes. It finds that the more uncertain a voter is about the candidates, the more likely he or she is to vote for the incumbent or a better-known candidate. This clearly reflects voters' risk-averse attitudes that reward the candidate with greater certainty, all other things held constant. Heterogeneity in voter uncertainty and its electoral consequences, therefore, have important implications for candidates' strategies. The findings suggest that the voter heterogeneity leads candidates' equilibrium strategies and campaign tactics to be inconsistent with those that spatial analysts have normally proposed.