Determinants of Persistence in Postsecondary Education,: An Analysis of the National High School Longitudinal Study of 2009

Abstract

Objectives As postsecondary enrollment rates have continued to increase (Digest of Education Statistics, n.d.), college student attrition rates have remained relatively consistent. More in-depth research on college persistence is important to understand the phenomenon and provide policy implications. This report investigates students’ persistence in postsecondary education and examines influential factors utilizing the High School Longitudinal Study of 2009 (HSLS:09) dataset.

Data and Method HSLS:09 is a nationally representative longitudinal study of 23,000+ 9th graders in 2009, with a first follow-up in 2012 and a second follow-up in 2016. We conduct descriptive analysis and regression models to analyze the trend of persistence in different groups and examine the impact of the possible determinants. We create a binomial logit model to measure persistence from the data set against three key variables which have been shown to predict educational attainment throughout literature: student ability (math skill), student attendance, and parental expectations. Of the three variables, our model only finds statistical significance with student ability. Ability is shown to have the strongest impact on persistence, followed by being female, having both biological parents in the household and attending private school. Having kids and being married or in a live-in relationship are shown to have the strongest negative impact on persistence. We include a variety of control variables to test the effects of the three variables and also to see how the control variables could affect persistence rate. The control variables, which are at individual, family and school levels, consist of: sex, race, family income, mother’s education, household composition, whether high school is public, whether students get married or have children in college. More specifically, subgroup regressions by sex are conducted to explore the potential different impact of these explanatory variables on males and females separately.