Logit analysis of graduate student retention

A logit model is developed to predict the retention of graduate students. The use of the BMDP4F procedure to develop a logit model for graduation student retention is outlined, and the model is applied to data from a major university. In this study, predicted retention rates for master's students were independent of age and sex, but were a function of race/visa status, registration status, and the interactions between academic division and registration status and between academic division and race/visa status. Predicted rates for doctoral students were independent of age and race/visa status, but were a function of registration status, academic division, and the interactions between academic division and registration status and between academic division and sex.