Key points are not available for this paper at this time.
Using 202 subjects, this study develops and tests a configurational approach to the contingent consistency concept of the attitude-behavior relationship. Its theoretical position is that neither attitude, nor social situational variables adequately predict behavior when treated separately. When combined in a linear model based on interaction between variables, behavioral predictions are improved considerably. Attitude toward the legalization of marijuana was measured along with perception of the position of peers and family. At a later date subjects were observed in an experimental situation where they were asked to vote to legalize marijuana. A multiple regression solution showed that the independent effects of the three variables did not clearly predict behavior, their additive effects did somewhat better, and the interactions (configurations) led to the clearest predictions. These results supported a theoretical orientation similar to field theory. It should be noted that only one attitude object was used and only two situational variables were considered along with attitude. Nonetheless, the results strongly suggest that a configurational approach promises to unravel the attitude-behavior inconsistency problem generated in recent years. Therefore, the approach deserves further development and testing as a specific model of the contingent consistency relationship between attitudes and action.
Acock et al. (Fri,) studied this question.