Generalizability of Factor Structures Underlying Student Ratings of Instruction.

The generalizability of factor structures of student ratings of instruction based on instructors' individual differences was examined. The subjects were instructors from the humanities, social science, and mathematics divisions who had had their courses evaluated at least twice using the same evaluation questionnaire. The data from the three divisions were factor analyzed and the resulting factor structures compared using Kaiser's procedure. Only one factor ("defines student responsibilities") was found to be stable within the across divisions. The results were explained in terms of the distinction between within-class covariation and between-class covariation. (Author)