The stability of teacher performance in the same course over time

Traditional reliability studies report the stability of teaching performance, as measured by student course/instructor evaluations, as ranging from .40 to .90, depending upon the study. Recently, generalizability theory has been applied to these rating data to more specifically pinpoint sources of error in student course evaluations. The purpose of the present study is to (1) familiarize the reader with some of the work that has been done with student rating data using generalizability theory and (2) to illustrate the use of the theory on a problem that in the past has been examined using traditional reliability estimates. Specifically, the paper addresses the stability of teacher performance across time in the same course.