Determinants of Student Evaluations of Global Measures of Instructor and Course Value

Abstract

Using structural equation modeling (LISREL) with a sample of business students, this study addresses important issues with respect to student evaluations. After reviewing past research and using exploratory factor analysis, confirmatory factor analysis suggests the existence of five latent variables in student evaluations: organization, workload/difficulty, expected/fairness of grading, instructor liking/concern, and perceived learning. Next, adding structural paths, these variables are used to model the formation of student evaluations. The models developed demonstrate that student evaluations may lack discriminant validity, the extent to which a measure does not correlate with other constructs it is not supposed to measure (e.g., expected/fairness of grading does have a large impact on ratings of teaching ability). Moreover, the use of student evaluations is compounded by the common usage of global measures in student evaluation practice. Here, the use of overall course value and instructor rating are contrasted. The results are disparate, raising questions about the use of summative measures (i.e., of what should they be composed, if used at all).