Quercus Engagement as a Predictor of First-Year University Student Dropout Risk: A Logistic Regression Analysis

Student retention is crucial for universities, impacting both institutional success and student outcomes. This study examines whether first-year engagement on Quercus, the University of Toronto’s online learning management system, predicts second-year dropout risk. We analyzed data from 3,298 students in the 2022– 2023 cohort, using logistic regression models to assess standardized cumulative Quercus visit counts over four observation windows (Weeks 1-4, 1-8, 1-12, and 1-15). Higher Quercus engagement was consistently associated with lower dropout probability, with the first four weeks proving sufficiently predictive. However, limited model sensitivity suggests engagement data alone may not accurately identify at-risk students. These findings highlight the potential of early Quercus disengagement as a warning sign while underscoring the need for advanced modeling techniques, such as machine learning or time-series analysis, and additional behavioral factors, such as course performance, extracurricular involvement, and support-seeking behaviors, to enhance predictive accuracy and support timely interventions. Click project repo to view more.