Understanding Persistence of At-Risk Students in Higher Education Enrollment Management Using Multiple Linear Regression and Network Analysis

Abstract

This article presents research on persistence among at-risk students using network analysis following multiple linear regression (MLR). Data on a population of enrolled undergraduate students at an urban-serving university over several years (P = 35,239) is tested using multiple linear regression. Variables interacting at different dimensions of the model are analyzed using IBM SPSS Neural Networks (2017) and Cytoscape (2018) to show network linkages. While variables may not be significant in MLR models where anomalies are ruled out, network analysis takes these anomalies into account and reveals complex layers of interactions between and among variables. Findings show that loans of any kind contribute to attrition while financial aid of any kind contributes to persistence and offsets attrition from loans.