Determinants and probability prediction of college student retention: new evidence from the Probit model

In this study, we adopt the Probit model and employ data on nine new freshmen cohorts at a public regional university to examine predictive factors for students' retention and measure projected probability of retention for an average college student. Results suggest four main findings: 1) high school GPA and class rank are significant positive retention predictors while ACT scores are not; 2) graduating from a larger high school indicates greater likelihood of retention; 3) controlling other factors, being female indicates lower likelihood of retention; 4) programmes including orientation or remedial English courses, on-campus jobs, and on-campus residence positively impact retention. In addition, we discuss implications that connect retention with grade inflation. Grade inflation will have negative impacts on institutions and students, not only in the higher education market but also in both the job and knowledge markets in the long run.