Predicting Student Retention in STEM Majors

• Activity Number: 256
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Education
• Author(s): Andrew Sage* and Dan Nettleton and Cinzia Cervato and Craig Ogilvie
• Companies: Iowa State University and Iowa State University and Iowa State University and Iowa State University
• Keywords: STEM ; random ; forest

Abstract

Approximately ten percent of Iowa State undergraduate students who plan to major in science, technology, engineering, or mathematics (STEM), switch to non-STEM majors within their first year at ISU. Early identification of at-risk students might enable departments to provide interventions aimed toward helping these students succeed in STEM. In this research, we use random forests to estimate the probability of individual students leaving STEM based on information available to the university early in the student's first semester of study. We also identify variables that are strong predictors of a student leaving STEM. Potential explanatory variables include high school academic performance, test scores, career goals and interests, ISU course placements, and participation in activities such as learning communities. The situation is complicated by the presence of missing values. Several imputation techniques are considered and the impact of imputation on variable importance measures is assessed.