Forecasting Teaching Evaluations for Student, Teacher, and Relational Components: Predicting more of the professors more of the time
College of Liberal Arts and Sciences
Social and Behavioral Sciences
In applied settings, psychological measurement often relies upon human judgment and forecasting these judgments is an important research focus. Judgments are typically composed of a blend of three influences: raters, the targets who are rated, and the relationships among raters and targets. Nonetheless, most research focuses on only one of these influences, likely limiting predictive accuracy. We forecasted students evaluations of college lectures from students reactions to brief video trailers. We found that teaching evaluations were influenced strongly by three sources of variance: students (raters), targets (teachers), and relationships (student-teacher matches). We also found that students responses to the Teaching Trailers powerfully forecasted students responses to live lectures. Moreover, for student-teacher relational influences, teaching evaluations were linked to quiz performance. The professor who elicited unusually high teaching evaluations from a specific student, also elicited unusually high memory for lecture (r = .16*) on the low-stakes quiz. Positive affect in response to trailers was an excellent predictor of teaching evaluations (r = .94*). Thus, isolating three influences permitted greater predictive accuracy than focusing on a single influence. Implications for forecasting human judgments in general, as well as student teaching evaluations specifically were discussed.
The David A. Kenny Festschrift: Advances in Social Psychology
University of Connecticut University Events & Conference Services 438 Whitney Road Ext., Unit 1185 Storrs, CT
Gross, Jennifer and Lakey, Brian, "Forecasting Teaching Evaluations for Student, Teacher, and Relational Components: Predicting more of the professors more of the time" (2014). Faculty Scholarly Dissemination Grants. 902.