Abstract
Persons registered to vote in Seattle, Washington for the November, 1986 general election and a September, 1987 primary election were randomly assigned to treatments in two telephoneconducted experiments that sought to increase voter tumout. The experiments applied and extended a "self-prophecy” technique, in which respondents are asked simply to predict whether or not they will perform a target action. In the present studies, voting registrants were asked to predict whether or not they would vote in an election that was less than 48 hours away. This technique, which previously increased turnout in a small study done during the 1984 U.S. Presidential election, was again effective among moderate prior-turnout voters in the second of the present much larger experiments. The failure of the effect in Experiment 1 was plausibly a ceiling effect due to very high turnout for a U.S. Senate contest in the 1986 election. Successful applications of the self· prophecy technique are facilitated by social desirability of the target action (which leads subjects to predict that they will perform it). However, social desirability of the target behavior is not a sufficient condition for the effect, as indicated by an unexpected nonoccurrence of the effect among low prior-tumout voters in Experiment 2.