Reviews of Important Papers on Forecasting,
1985-1995 Reviews
Review of:

Anthony G. Greenwald, C. G. Carnot, R. Beach, and B. Young (1987), "Increasing voting behavior by asking people if they expect to vote," Journal of Applied Psychology, 72, 315-318.

[Review written with J. Scott Armstrong]

Previous studies have shown that if you ask people to predict their behavior, they tend to respond in a socially approved manner. They want to appear in a favorable light, so they report that they would act in a socially responsible way. These predictions about their behavior influence their later behavior; people live up to their predictions. Steven J. Sherman, who did one of the early studies on this, referred to it as a 'self-erasing error bf prediction'. I think it might also fall under the term "self-fulfilling prophecy." Greenwald et at. extend this by looking at voting behavior. A sample was selected of 62 students who had not yet registered. They were contacted by phone. If they did not know how to register, they were told how to do so. A randomly selected half of this sample was then asked if they intended to register to vote in the Reagan-Mondale election which was to take place the next month. Those who did intend to register were then asked for the most important single reason that they would vote, a procedure that was based on recomrnendations from prior research. Of this "prediction" sample (n = 32), 68.8% said that they would register. As might be expected, this was significantly higher than the 40.6% who actually registered. Most importantly, the 40.6% who actually registered was higher than the 33.3% who were not asked to make a prediction. While this increase in registration is substantial and is in the correct direction, it is not statistically significant. A second experiment asked about voting. The design was similar. Here, 100% of the prediction sample said that they would vote, and 87.5% actually did so. This actual voting exceeded the 64.3% for the non-prediction sample. The increase in voting was statistically significant at p <0.05.

Thus, the two experiments add support to a finding with important implications for managers. For example, self-predictions can be used to increase socially responsible behavior within a firm on issues such as "Would you cheat on an expense account?" "Would you accept a bribe?" "Would you cheat a customer?" (If they do not predict socially responsible behavior, then you know you have a problem). Universities can use it also. "Would you cheat on an exam?" or "Would you plagiarize?" Or it could be used for individual behavior as, for example, in trying to change behavior with respect to the AIDS epidemic: "You are attracted to someone. Would you use a condom?" Self-predictions should be encouraged for situations involving social responsibility. They are likely to be self-fulfilling predictions.