Abstract
Correlations between student ratings of the same course taught by the same instructor on two different occasions (n = 341 pairs of courses) were high (mean r = .71), but were lower than the reliabilities of the ratings (mean r = .93); much reliable variance was unique to a particular course offering. This study investigates factors that might explain this unique variance. For each pair of courses, the more favorably evaluated tended to be: (1) the one in which students expected higher grades (and presumably learned more); (2) the one which students perceived to require the most work; and, (3) the one which was taught after the instructor had already taught the course at least once before (and presumably improved as a consequence of this experience). These findings are not consistent with the hypothesis that these background characteristics bias student ratings, and they argue for alternative explanations.
Get full access to this article
View all access options for this article.