Analyzing dropout rates of B.Sc students

Abstract

Survival analysis is a useful statistical technique for analyzing failure time data. It overcomes the limitations of cross-sectional analysis and convention regression analysis. This study proposes three statistical models that analyze the failure time when university students withdraw from B.Sc courses. The sample comprises all the 91 students who commenced their four-year studies with the Faculty of Science in 2002. The first approach uses the Kaplan-Meier product limit method for estimating the survival functions under non-informative censoring. The second approach uses the Cox regression model, which involves the assumption of proportional hazard functions. The third approach uses a parametric model to estimate the hazard function using an appropriate distribution. The aim of the study is to fit survival models that predict the probabilities of retention and dropout of B.Sc students within each study area using the facilities of SPSS and STATA.