Exploring Important Motivation Drivers for Completion of Massive Open Online Courses by Bayesian Hierarchical Logistic Regression Model

Abstract

Massive Open Online Courses (MOOCs) have provided over 9 thousand high-quality online courses to 81 million learners. MOOCs have brought about a revolution in education by eliminating the geographic and time constraints of traditional classrooms. It is also well-known that high dropout rates come with such massive user numbers. The retention rate of MOOCs may differ according to the various kinds of motivations that influence learners to finish courses. Few studies have taken into account the effect of user motivations on MOOCs' retention rates. This research involved comparisons between the effects of different variables on motivation as well as administrative variables in MOOCs by using Bayesian hierarchical logistic regression models. A Bayesian model allows explanatory variables that are not necessarily independent and treats missing values in variables within its method. This research also uses data from many course disciplines (instead of courses from the same discipline) and builds models for general data. The results show that hierarchical models performed better than non-hierarchical models, identifying the combinations of administrative independent variables and motivations that can contribute to the improvement of future architecture of MOOCs. Such improvements may help motivate targeting learners to actively engage in related courses on promotion.