ABSTRACT Attitude theory and opinion polls have become an important part of the social fabric. They affect the campaign strategy and tactics adopted by political candidates, the marketing strategy and advertising campaigns of Fortunate 500 companies, the programming development and strategy in mass media, and strategic planning and program development in public education. In theory, opinion polls provide a means for the public to feedback and shape institutions that govern their lives. This research is based on the premise that this theory is incomplete and, consequently, decisions based on current attitude theory and surveys can be grossly (although nonobviously) misleading. Specifically, contemporary attitude theory and opinion polls are based on the premise that people know and will report their evaluative dispositions toward attitude objects (e.g., candidates, products, educational options). However, there is now evidence from various disciplines that evaluative dispositions are represented and activated in more complex ways than accommodated by current theory and measures, and that contemporary bipolar (e.g., favor/oppose, like/dislike) self-report scales can potentially mask important differences in people's evaluative dispositions. A quantitative theory is proposed that subsumes existing attitude theory and research while calling attention to specific areas in which prior work can be quite misleading. A systematic program of research will contrast the predictions made by contemporary theory of attitudes and the proposed theory of "evaluative space." For instance, individuals who have no attitude toward or feel neutrally about a topic are indistinguishable using traditional opinion questions from individuals who are strongly conflicted about (e.i., who possess ambivalent attitudes toward) the topic. Attitudinal ambivalence results when both positive and negative evaluative processes are coactivated and, as such, reflects a special case of behavioral conflict. Research will investigate the determinants, mediation, and behavioral consequences of attitude ambivalence and conflict, and will integrate this and traditional attitude theory and research within the more general theoretical framework of evaluative space. In view of the ubiguity of evaluative processes in behavior, illumination of their dynamics and the underlying evaluative mechanisms should have important ramifications for attitude theory and research, and broad significance for basic theory in psychology generally.
