A model of student retention in learning communities

A model of student retention was developed from industrial and educational research on voluntary departure. Archival data were used from a national survey, joined with actual academic performance and persistence data from full-time freshmen in a learning community at a mid-sized private university in the mid-Atlantic. Exploratory and confirmatory factor analyses were used to develop the constructs of social and academic integration, academic services, satisfaction, institutional commitment, and intent to persist. Demographics and ability were controlled to reduce effects of non-random assignment. A hypothesized persistence model derived from the literature was tested for fit and path significance. Post hoc, a revised model was developed which excluded non-meaningful demographic and ability variables and pruned non-significant paths between constructs.; The resulting model explained 72% of variance in intentions to persist and 19% of retention behavior. Satisfaction mediated the effects of social and academic integration on intentions to persist. While the direct effect of social integration on satisfaction was stronger, the total effect of academic integration was about the same on both satisfaction and intention to persist. High school grade point average (GPA) was the only individual characteristic to be meaningfully predictive in the model and college GPA was not a significant predictor of intention to persist. Institutional commitment as predicted by academic integration was not a meaningful predictor of intention to persist. When retention behavior was added to the model, student intentions to persist mediated the effects of all other model constructs.; The final model was a better fit for men. While academic and social integration equally affected male intentions to persist, social integration was more meaningful to Models of female persistence intentions. The model was also a better for students in seminars. Social integration had a stronger effect on persistence intentions in seminars, while in linked courses academic integration had the stronger effect.