Increasing voting behavior by asking people if they expect to vote.

In two studies, students contacted by telephone were asked to predict whether they would perform a particular behavior (registering to vote or voting, respectively) in the next Tew days. The proportion who predicted that they would do these socially desirable behaviors exceeded the proportion of control subjects wh o performed the behavior without first being aske d to predict whether the y would . Further, in the voting study these errors of overprediction were in the sense describe d by S. J. Sherman (Journal o f Personality and Social Psychology, 1980,39). That is, subjects who were asked to predict whether they would vote—all of whom predicted that they would—actually did vote with substantially greater probability than did the no-prediction control subjects. (Actual voting was verified by consulting official voter rolls.) Asking people to predict whether the y will perform a socially desirable action appears to increase their probability of performing the action. When making predictions about their own behavior, people tend to present themselves favorably; they predict that they will do what appears to be proper or good behavior. However, when given the opportunit y to act , a person' s likelihood of performing a socially desirable action may b e reduced by factors such as the action's time and energy costs, the availability of compelling alternatives, and missed opportunity through not responding promptly. Sherman (1980) showed that asking people to predict their actions does more than jus t reveal a tendency toward favorable self-presentations; the probability of the predicted action is affected. Once subjects have made a prediction, their behavior is likely to confirm that prediction. In one of Sherman's experiments, subjects who were asked to predict whether they would agree t o work 3 hour s t o collect money for the American Cancer Society (49% said they would) were much more likely (31%) to agree with a later request to do so than were those who were never asked to predict their behavior (4%). Thus, apparent errors in prediction are self-erasing (Sherman, 1980). Once a person predicts an action, that action is likely to occur, even when the initial prediction is an apparent gross overestimate of the likelihood of performance. In interpreting this finding, Sherman suggested that making a prediction produces a performance-facilitating cognitive representation in which the person