The Impact of Small Learning Group Composition on Drop-Out Rate and Learning Performance in a MOOC

A commonly known and widely studied problem of massive open online courses (MOOCs) is the high drop-out rate of students. In this paper we propose and analyze the composition of small learning groups as a solution to this problem. In an experiment, we composed such small learning groups in a MOOC context using two methods: Random grouping and grouping by an algorithm that considers selected student criteria. Further, a flipped classroom course was conducted on-campus with a local student group using the MOOC. We compared all three approaches to a control condition using two measures: Drop-out rate and learning performance. The empirical results give an indication, yet no hard evidence, that small groups might reduce student drop-out rates. *