Logo image
Back
Characterizing instructional practices with cluster analysis; an exercise in reducing uncertainty
Abstract

Characterizing instructional practices with cluster analysis; an exercise in reducing uncertainty

Jordan Harshman, Marilyne Stains, Megan Barker, Stephanie Chasteen, Renee Cole, Sue Ellen DeChenne-Peters, M. Kevin Eagan, Joan Esson, Jennifer Knight, Frank A. Laski, …
Abstracts with programs - Geological Society of America, Vol.52(6)
Geological Society of America, 2020 annual meeting; GSA 2020 connects online
10/2020
DOI: 10.1130/abs/2020AM-356496

View Online

Abstract

Researchers have been interested in characterizing instructional practices for a variety of reasons. These classroom observations have become a go-to method for those in discipline-based education research. Because of the variety of instructional practices, researchers have struggled to generally characterize classroom practices. Throughout this presentation, it will be demonstrated how the Classroom Observation Protocol for Undergraduate Science (COPUS), in tandem with cluster analysis, was used to characterize the teaching practices of over 2,000 classroom observations. Seven instructional profiles were identified and later reduced to three: Didactic, Interactive Lecture, and Student-Centered. Clustering on variables that show such wide distributions comes at the cost of certainty. Therefore, this talk focuses on the issue of reducing uncertainty in these classifications despite the large variety demonstrated. This was done through model-based clustering algorithms that allow for specific fit parameters as well as probabilistic profile assignments that serve as evidence to support the final classifications. Lessons learned for researchers using COPUS and/or cluster analysis algorithms will be provided.
 Miscellaneous

Details

Metrics

45 Record Views
Logo image