Analysis of attrition and retention rates using the generalized linear model

Abstract

Improving the progression rates of students and reducing the numbers of students dropping out from institutions of higher education are critical to get maximum return for financial subsidies received or private fees paid, as well as being key components for producing skilled workers within the developing economy. Institutions of higher education in South Africa are accordingly grappling with finding a delicate balance between access, equity, redress and quality. A study of attrition is a sensitive, yet essential issue for university planning offices. An appropriate modelling approach is essential for identifying factors that contribute to attrition. This study presents two models for attrition, with slow and fast drifts of attrition, as students progress from year to year, with constant and varying dampening effects. The fast dampening model has the property of relative risk, whilst the slow drifting model has the property of odds ratio. The effect of faculty, gender, race and entry batch year, on the progression and attrition rates was examined in the study. The results of the analysis show that the first year attrition rate of White students is higher than that of the other race groups, whilst the retention and graduation rates of White students, is much greater than that of the Black and Indian students, from second year onwards. Throughout the three-year study period, the attrition rates of female students was found to be consistently lower than the corresponding attrition rates of male students.