Designing Feasible Optimal Treatment Regimes for Personalized Education

• In Event: Innovative Statistical Methods and Computational Tools for Complex Research Designs

Abstract

Current math course-taking plans for high school students are often not tailored to support individual success in mathematics. In this study, we apply optimal treatment regime (OTR) methods to develop personalized course-taking plans. However, these methods may yield recommendations that are difficult to implement in real-world settings. To address this, we incorporate two feasibility constraints—propensity-score-based and a priori constraints—into the OTR methods to ensure the recommendations are realistic. We conduct simulation studies under two OTR methods, Q-learning and Targeted Maximum Likelihood Estimation (TMLE), with and without these constraints. We also demonstrate the effectiveness of the proposed feasible OTR methods using data from the High School Longitudinal Study of 2009, and conclude with guidance on selecting appropriate feasibility constraints.

Authors

• Yuxuan Li, Teachers College, Columbia University

• Chenguang Pan, Teachers College, Columbia University

• Youmi Suk, Teachers College, Columbia University