In the analysis of data from a college sample, the validity of the Authoritarian Personality Theory of racial and ethnic prejudice is questioned, and thus, through this, the psychologistic approach to the study of social attitudes. A rigorously sociologistic approach is shown to be a better predictor of variations in the data than the Authoritarian thesis, even for the covariations between F-scale scores and scores on two different prejudice scales. The importance of differential learning for understanding variations in racial prejudice is stressed.