Positional discrepancy, psychological discrepancy, and attitude change: Experimental tests of some mathematical models¹

Abstract

Conflicting models of the attitude change process were tested in two experiments. The simple linear balance model predicts that the amount of attitude change induced by a message is proportional to the discrepancy between the respondent's original position and the position advocated. The psychological‐discrepancy‐discounting model assumes that as messages become more psychologically discrepant from one's position, they lose their effectiveness in a nonlinear manner. This model assumes that the psychological discrepancy of a message is influenced by the other messages accompanying it (e.g., the presence of a more extreme message can make a given message seem less psychologically discrepant than it would otherwise be). This model also suggests that placing an extreme message before a moderate one will produce less psychological discrepancy and more attitude change than would the reverse order. In two studies (N = 193 and N = 114), subjects read one or two messages, each advocating either a 15 percent or a 50 percent increase in tuition (or in control conditions, read no message at all), and then indicated the tuition increase they favored. Subjects were also asked how psychologically discrepant they found each message. The results supported the predictions regarding the effect of other messages on the psychological discrepancy induced by a message. Where the message was accompanied by supportive arguments, the attitude change results supported the psychological‐discrepancy‐discounting model much better than the linear balance model.