Wamala, RobertOonyu, Joseph C.Ocaya, Bruno2022-12-062022-12-062012Wamala, R., Oonyu, J. C., & Ocaya, B. (2012). 4. A comparative assessment of selected approaches in modeling completion dynamics of graduate programs. African Statistical Journal Journal statistique africain, 58, 57.2233 2820https://nru.uncst.go.ug/handle/123456789/5905As the higher education landscape changes, factors related to student persistence in graduate programs are being examined by a growing number of researchers. Their investigations, however, can be problematic when: i) a considerable number of students have not completed the program at the time of data collection; ii) enrolment and completion figures are low; iii) a normal distribution of completion time is assumed; and iv) a detailed assessment of non-completion is required. A time-to-event approach in a Cox model, which uses enrolment time as censored for students who have not completed by the time of the study and makes no assumption about the distribution of completion time, solves the first three problems. A multinomial logistic, allowing for at least three outcomes of doctoral candidature, solves the fourth problem. This is illustrated by an analysis of administrative data for the 295 PhD students at Makerere University in the 2000 to 2005 enrolment cohorts. The total elapsed time from first enrolment to submission of final thesis copy was adopted as a measure of completion time. The findings underscore the suitability of these approaches for the analysis of education data with low enrolment and completion figures, a situation characteristic of doctoral studies in many African universities.enStudent success in graduate programsCompletion and non-completion of doctoral studiesMakerere UniversityArticle