Predicting Team Function Using Bayesian and Cognitive Diagnostic Modeling Approaches

Abstract

Team-based learning is commonly used in engineering introductory courses. As students of a team may be from vastly different backgrounds, academically and non-academically, it is important for faculty members to know what aid or hinder team success. The dataset that is used in this paper includes student personality inputs, self-and-peer-assessments of teamwork, and perceptions of teamwork outcomes. Using this information, we developed several Bayesian models that are able to predict if a team is working well. We also constructed and estimated Q-matrices which are crucial in explaining the relationship between latent traits and students' characteristics in cognitive diagnostic models. The prediction and diagnostic models are able to help faculty members and instructors to gain insights into finding ways to separate students into teams more effectively so that students have a positive team-based learning experience.