Pre-Election Polling and the Rational Voter: Evidence from State Panel Data (1986-1998)

This study examines the role of pre-election perceptions of race closeness, by way of newspaper polls, in motivating citizens to vote on a specific contest while in the voting booth. Results from an error components model (random effects), employing state panel data of concurrent gubernatorial and Senate elections for the period 1986 to 1998, fail to bolster the rational voter hypothesis that perceived closeness matters. The extent to which pre-election perception matters is shown to depend directly on how one measures the likelihood of a close contest. In contrast, few long-standing variables, inherent to the voting literature, have any impact on "within-booth" voting behavior. The majority of within-booth abstention is left unexplained; furthering the notion of Matsusaka and Palda (1999) that the act of voting, indeed, may be random.