John will defend his thesis, “Understanding University Student Pathways Towards Graduation with Machine Learning and Institutional Data” on Friday September 24, 2020 at 15:30 Oslo Time. Connection information is available on the University of Oslo Physics website.

Title: A new framework for evaluating statistical models in physics education research (PER)

Abstract: Across education there has been an increased focus on the development, critique, and evaluation of statistical methods and data usage due recently created very large data sets and machine learning. In physics education research (PER) this has recently been examined through the 2019 Physical Review PER Focused Collection examining quantitative methods in PER. Quantitative PER has provided strong arguments for reforming classrooms via interactive engagement, demonstrated that students often move away from scientist-like attitudes due to science education, and has injected robust assessment into the physics classroom via concept inventories. This presentation examines the impact that machine learning will have on physics education research and presents a new framework for evaluating statistical models in PER. This paper then demonstrates the utility of this evaluation framework through simulations, analysis of survey data, and analysis of student pathways both in a physics major and across degree programs.