As expected, students involved in a study of college impact did, in fact, become more "liberal" politically and less "traditional" in their religious views between their entry into college and the beginning of their sophomore year. Data collected to test a model of student attitude change indicated that the internal consistency of variable sets included in a system of college impact did differ in predicted directions. Further, the psychological and social factors that characterize the impact system show smaller interrelationships for students who change their political attitudes than for those whose attitudes remain stable. The social factor is the stronger set of predictor variables. The present data point out, however, the futility of generalizing across college systems as if they were identical regarding either the students they attract or the processes of influence they exert on those students once they are matriculated. (Author/TS)