Who Will Succeed and Who Will Struggle? Predicting Early College Success with Indiana’s Student Information System
College success and career readiness have become major goals of education reform. Toward this end, Indiana policymakers have undertaken multiple efforts to prepare students for college. This study supports those efforts by describing the early college success of Indiana students, identifying measures in the state longitudinal data system that predict early college success, and examining the usefulness of those predictors.
Using data from the Indiana Student Information System, the state’s longitudinal data base, this study examined the early college success of Indiana’s 2010 high school graduates who entered an Indiana two- or four-year public college in fall 2010. Because there is no widely accepted single indicator of early college success, the study adopted three commonly used indicators—enrolling in only nonremedial courses in the first semester, completing all attempted credits in the first semester, and persisting to a second year of college—plus a measure consisting of a composite of all three indicators.
Half of Indiana 2010 high school graduates who enrolled in a public Indiana college were successful by all indicators of success. College success differed by student demographic and academic characteristics, by the type of college a student first entered, and by the indicator of college success used. Academic preparation in high school predicted all indicators of college success, and student absences in high school predicted two individual indicators of college success and a composite of college success indicators. While statistical relationships were found, the predictors collectively only predicted less than 35 percent of the variance. The predictors from this study can be used to identify students who will likely struggle in college, but there will likely be false positive (and false negative) identifications. Additional research is needed to identify other predictors—possibly non-cognitive predictors—that can improve the accuracy of the identification models.
