An application of event history modeling to assessing student dropout behavior: A national data approach

This study investigated the dropout behavior of college students. Previous attrition studies have typically used structural path modeling to investigate student departure. While path models have been shown to be valid in describing students' dropout behavior, they lack a more practical application. For instance, they do not explain when dropouts occur. Although longitudinal data are often used in structural path models of student departure, generally the role of time has not been adequately addressed. Also, path models and other analytic techniques do not consider that included independent variables may have effects on student departure that vary over time. The study presented below uses event history modeling to examine the timing of dropout as well as how the effects of independent variables may vary over time.;The data used in this study is the Beginning Postsecondary Students Longitudinal Study Second Follow-up sponsored by the National Center for Education Statistics. The effective sample used includes 3,450 students who attended four-year public and private institutions between 1989 and 1994. Different types of event history methods were used to examine the probability of dropout and the time-dependency of the variables hypothesized to affect student dropout behavior.;The results demonstrate that dropout behavior varies over time. The highest risk period for dropout is the spring semester in the freshman year. Also, students from lower income families are, in general, more likely to drop out than their counterparts. Interestingly, this effect is most pronounced in the second and third years of enrollment. The effect of mother's educational background is also significant in the second and third years. Finally, the results suggest that students who receive aid have lower dropout rates than non-aided students, and over time dropout rates vary depending on the amount of aid a student receives.