Campaign Learning and the Fundamentals Reconsidered

Abstract

The contrast between the predictability of presidential elections and the variability of early polls has come to be viewed as evidence of campaign learning. We argue that unmotivated survey respondents offering minimally acceptable answers (i.e., satisficing) offers an additional explanation for the classic conundrum of why the polls vary when the election outcome is predictable. The analysis relies on data from the National Annenberg Election Survey, a natural experiment that results from California's election laws, and the 2000 ANES survey mode experiment. The results support the claim that respondents' motivation to engage the survey question, not the information provided by the campaign, is the most important determinant of whether vote intentions reflect the "correctly" weighted fundamentals. We conclude by discussing the implications of this finding for both survey-based and experimental studies of campaign effects.

Keywords: polls, elections, fundamentals, vote choice, satisficing