WHAT KINDS OF ATTITUDE MEASURES ARE PREDICTIVE OF BEHAVIOR?

In: Public opinion quarterly: journal of the American Association for Public Opinion Research, Band 35, Heft 3, S. 327-334

Abstract

An analysis of the conditions common to 3 types of situations in which the r between verbally expressed att's about behavior & the behavior itself are likely to be high: voter turnout in elections; movie attendance; & brand preference. 3 corresponding studies are discussed: Paul K. Perry (see SA A1913); Paul K. Perry, 'Marketing and Attitude Research Applied to Motion Pictures,' paper presented at the Internat'l Gallup Conference, New Delhi, Mar 26, 1968, privately printed; & Irving Crespi, 'Toward a Definition of 'Attitude' ' in Conrad Hill, Ed, INSIGHTS INTO APPLIED BEHAVIORAL RESEARCH (Kingston, RI: U of Rhode Island, 1969). The concern was with the predictive power of the highly specific attitude measurements obtained in these 3 studies. It is felt that improving behavioral prediction requires test items having stimulus properties very similar to those present in the actual situation. In all 3 studies, att's were treated as multidimensional, with no 1 dimension acting as a 'predisposing' determinant of behavior. Instead of conceptualizing act's as general predisposing variables which are expected to have common effects in a wide variety of situations, att's were treated as highly specific combinations of beliefs, preferences, & intentions, each held with varying degrees of intensity. The validity of this approach is affirmed. A multidimensional, functional conceptualization of att's is far more productive empirically than the mechanistic view that att's are predispositions, that is, intervening variables in a sort of complicated pattern of conditioned behavior. Att's originate, develop, stabilize, change, & atrophy in a dynamic process of role playing. M. Maxfield.