Distinguishing Biases From Genuine Differences in Student Evaluation of Teaching

• In Event: Marriott Roundtable Session 20
  In Roundtable Session: Understanding and Examining Student Evaluations of Faculty Teaching

Abstract

Those concerned with the validity of teaching evaluation may rightfully believe that teaching quality should not differ as a function of instructor’s characteristics such as gender or race. A difference favoring one type of instructor (e.g. female instructors) would indicate biased scores. However, only an unfair difference, a difference explained by irrelevant aspects of the evaluation, would compromise the results, a point omitted in Student Evaluation of Teaching (SET) literature. The proposal defines, justifies and illustrates test fairness analysis using item response models and differential item functioning with data collected from Education graduate students in North America. We conclude about how test fairness contributes to the validity of SET and the study of biasing factors.

Authors

• Edgar Valencia, Pontificia Universidad Catolica de Chile

• María Inés Godoy, University of Chile/DEMRE