Self-Perception and Attitude-Behavior Consistency

An experiment tested the hypothesis that the degree to which an attitude is inferred from past behaviors toward an object is a determinant of attitude-behavior consistency. Subjects participated in a two-session experiment. During the first session, subjects' attitudes toward religion were assessed under one of two conditions. Control subjects' attitudes were measured before they completed a questionnaire wherein they indicated which of a list of behaviors pertaining to religion they had performed in the past. Self-Perception subjects completed the same two sets of items, but in the reverse order; thus, they were provided with an extensive opportunity to review and evaluate their past behaviors before indicating their attitudes toward religion. Approximately one month later, all subjects returned for the second session, during which three measures of their behaviors pertaining to religion since the first session were obtained. On two of the three measures, the predictive utility of religious attitudes was shown to be greater in the Self-Perception condition than in the Control condition.