Applying Hierarchical Non-Linear Modeling To Investigate The Influence Of High School Characteristics In Predicting First-Year College Retention

Abstract

Student retention in higher education is an incredibly important social and psychological phenomenon. The impact of student retention reaches across multiple domains, influencing individual students, state and local economies, and even national prestige and viability on a global scale. Although a great deal of research has been conducted examining the influence of both student and school characteristics on student retention, less is known about how the two variables interact with each other. The present dissertation is designed to examine this phenomenon by applying multi-level modeling to estimate how variables at the school level (such as graduating class size and district spending) interact with variables at the student level (such as high school GPA and financial concerns) to predict first-year retention in college. Psychosocial and academic data were collected from over 6,500 students across 950 schools and applied to construct a series of multi-level models to estimate retention. Clustering analysis was then applied to see if the schools could be grouped according to “type” rather than used individually. Results indicated that multi-level models could be used to predict student retention in higher education, and that the most influential predictors of retention were academic and financial in nature. Implications and future research are discussed.