An empirical investigation of the potential confounding variables in student evaluation of teaching

Prior research in psychology and education has extensively investigated the propriety of using student evaluation of teaching (SET) for administrative control purposes. This literature is in general agreement that SET are susceptible to variables which affect the SET but do not relate to effective instruction. A recent American Accounting Association (AAA) -sponsored monograph by Williams, Tiller, Herring and Scheiner (1988) suggested that accounting education research explore the extent to which reported validity results in these literatures carry over to accounting. The objective of the present study is to investigate three major questions: (a) Do SET differ across upper-level undergraduate accounting courses?, (b) What potential confounding variables are most highly associated with SET?, and (c) Can a model which includes these variables be constructed so that administrators might use it to control for these variables? Our results indicate that significant associations exist between SET, course identity, and grading policy.