New Tools, Old Questions: Rethinking Success in Gateway Statistics Courses through Machine Learning and Theory

• In Event: AERA Poster Session Thursday 7:45am
  In Poster Session: Poster Session: Statistical Theory, Quantitative Methods and Applications

Abstract

This study investigates determinants of academic success in an introductory psychology statistics course using ensemble machine learning methods. Survey data from 186 undergraduates at a Hispanic-Serving Institution were linked to institutional grade records. Variables included statistics anxiety, statistics self-efficacy, demographics, and perceptions of relevance. Using ensemble techniques combining Least Absolute Shrinkage and Selection Operator (LASSO), random forest, and stepwise regression, the analysis identified statistics self-efficacy and anxiety as the most impactful predictors of course grades. Notable group differences emerged across gender and race. These findings highlight the utility of machine learning for educational research and underscore the role of psychological factors in statistics performance, with implications for instruction and student support.

Author

• Sunny Nguyet Le, California State University - Fullerton