Predictive Modelling for High Impact Active Learning Classrooms

A large body of research over the past several decades has shown that students learn more and more equitably in classrooms that are not entirely lecture-based. Non-lecture activities are collectively referred to as active learning, but this term lacks definition, and little research has been done on which types and combinations of active learning methods (e.g., group worksheets, clicker questions) are most effective. Using a dataset of 61 previously published undergraduate STEM classes from a range of fields and institutions, we develop a predictive model that maps time spent on specific classroom activities to student learning outcomes. In this talk, I will discuss the model-indicated types of classes that result in high impact student learning and the implications of these results for undergraduate STEM instructors.