Student ratings of instruction: A causal analysis of process variables

This manuscript applies structural equation modeling, a statistical technique that is used frequently for examining causal-effect relationships between latent constructs, to a widely used student rating of instruction form(SIRS). As such, the causal relationships between latent constructs depicted in the form are examined for evidence of nomological validity. The results indicate several interesting relationships among the process variables (student-instructor interaction, course organization and course demand) and outcome variables (instructor involvement and student interest) commonly used for evaluation of instruction. The study clearly supports the notion that process variables have significant effect on the outcome variables. This in turn raises some fundamental questions regarding the potential usages of such student evaluations. For example, should they be viewed as diagnostic tools for evaluating teaching process or evaluations of the teaching process and/or the teacher?