Will they Come? Modeling Matriculation Decisions for Admitted Applicants at the University of Arizona

Abstract

This study investigates factors influencing matriculation decisions for freshman applicants in the College of Agriculture and Life Sciences (CALS) at the University of Arizona. Two different modeling approaches are used on a five-year cross-sectional sample of applicants. Consistent with previous literature, a parametric logistic regression is specified to estimate the probability that a freshman applicant will matriculate in CALS. Additionally, this study also uses non-parametric gradient boosting methods to predict whether an applicant will matriculate. As a byproduct of using two different techniques to model matriculation decisions, an additional academic interest is to see how these two distinct approaches compare in terms of explanation and predictive capabilities. The results show that students who apply early and applicants with high standardized test scores are significantly less likely to matriculate. Moreover, applicants who attend campus tours, honor students, and students from high schools with many applicants are more likely to matriculate.