The diversity in reasons that students have for enrolling in massive open online courses (MOOCs) is an often-overlooked aspect while modeling learners’ behaviors in MOOCs. Using survey data from 11,202 students in five MOOCs spanning different academic disciplines, this study evaluates the reasons that students enrolled in MOOCs, using an unsupervised learning method, Latent Dirichlet Allocation (LDA). After fitting an LDA model, we used correspondence analysis to understand whether these reasons were general, and could be invoked across the five MOOCs, or whether the reasons were course-specific. Furthermore, log-linear models were employed to understand the relations between the reasons students enrolled, the course they took, and their background characteristics. We found that students enrolled for many different reasons, and that their age was statistically related to the reasons they gave for taking a MOOC, but their gender was not. The paper concludes with a discussion of how instructors and course designers can use this information when creating new—or redesigning existing—MOOCs.

Original languageEnglish (US)
StatePublished - 2018
Event11th International Conference on Educational Data Mining, EDM 2018 - Buffalo, United States
Duration: Jul 15 2018Jul 18 2018


Conference11th International Conference on Educational Data Mining, EDM 2018
Country/TerritoryUnited States


  • Informal education
  • MOOCs
  • Text mining

ASJC Scopus subject areas

  • Computer Science Applications
  • Information Systems

Cite this