Search

Using new robust methods for making causal inferences from survey data, I show that voters in U.S. Presidential elections are able to use cues and heuristics to act as if they are informed. More precisely, when voters are matched on baseline characteristics, voters who change information state vote no differently on election day than voters who do not. Before election day, however, voters who change information state do differ in their vote intentions when similarly matched. The election campaign appears to provide the necessary cues to make the differences vanish by election day. This evidence is consistent with a growing body of formal work which shows that voters are able to use cues, heuristics and other information shortcuts to act as if they are informed. Thus, much of the widespread worry on the part of democratic theorists and activists about the poor information state of voters is misplaced. The methods used are more robust than the traditional regression approach and are well suited to making causal inferences from observational data. The analysis relies on data from the National Election Studies particularly the 1972-1974-1976, 1980, and 1992-1994-1996 panels. [ABSTRACT FROM AUTHOR]