This project studies attitude change toward U.S. Presidents using the conceptual framework and the methodology of a mathematical theory of information integration. The subjects received various paragraphs of information about certain U.S. Presidents, and evaluated them on general statesmanship. This evaluation requires the subjects to combine or integrate the information in the various paragraphs into a unitary attitude. Integration theory makes some straightforward predictions about how the information is integrated. Remarkably good results were obtained. Experiment I provided an initial test of the parallelism prediction of integration theory. It may be concluded that the information paragraphs were integrated into an overall attitude by a simple algebraic rule. Experiment II provided a more extensive test of the parallelism prediction. The results eliminated the adding hypothesis and supported the averaging hypothesis. The averaging effect seems fairly conclusive and should resolve the theoretical controversy centered on averaging-versus-adding. A very important methodological product of this work is the set of president paragraphs. Favorableness values are given for each president included. These should be useful in the general study of attitude development. Finally, integration theory leads to a new view of the relation between attitude change and learning theory. (Author/SLD)