Models of Generalizability Theory in Analyzing Existing Faculty Evaluation Data.

In this article, we demonstrate the use of generalizability theory in analyzing existing faculty evaluation data. Three measurement conceptualizations representing different purposes of faculty evaluation were conceived to account for the existing data structure. Variance components associated with these conceptualizations were estimated from random samples taken from an existing faculty evaluation conducted in a university in the western United States. Within each of the 3 conceptualizations, 3 decision study considerations are presented together with the generalizability coefficient estimates. Practical implications for faculty evaluations are discussed.