Observed vs Predicted Persistence and Graduation

The April 2020 report explores differences between an undergraduate department’s observed second-fall persistence and six-year graduation rates and the department’s predicted rates, based on the demographics and high school GPA of the three most recent first-time, full-time (FTFT) cohorts. The intent of this analysis is to encourage conversation around policies and curriculum at the department level that could be associated with observed rates that differ from what would be predicted based on students’ application materials.

To generate department-specific predicted rates, IRP&E created separate by-college logistic regression models to obtain student-level predicted probabilities of persisting to second fall for first-year cohorts and of graduating by sixth summer for junior cohorts, controlling for a suite of student demographic characteristics. Generating predicted probabilities by colleges separately is necessary as student profiles differ between the colleges at CSU (e.g., first generation, Pell, and racially minoritized students comprise a larger share of some colleges more than others).

The report finds that the departments of Business intra-university, Construction Management, Design & Merchandising, Food Science & Nutrition, Human Development & Family Studies, and Economics have observed persistence and graduation rates higher than the predicted. In contrast, the departments of Health & Exercise Science, Art & Art History, Biology, Mathematics, Fish, Wildlife, & Conservation Biology, and Electrical & Computer Engineering have observed persistence and graduation rates lower than the predicted.

Notably, the departments of Human Development & Family Studies and Biology have higher than average proportions of at least two gap populations of interest, first generation, Pell, or racially minoritized, for both the FTFT and junior cohorts. Design & Merchandising, Health & Exercise Science, and Mathematics have higher than average proportions of at least two of these populations for the FTFT cohorts only.

The report’s Appendix provides more details on the college-specific logistic regression models as well as department-specific demographic statistics.