Using an integrated probabilistic clustering approach to detect student engagement across asynchronous and synchronous online discussions

Abstract

Online collaborative discussion (OCD) focuses on promoting individual knowledge inquiry and group knowledge construction through active peer interactions and communications. In practice, it is necessary to explore how different modes of OCD come into play, in which student engagement can function as an evaluating indicator. To identify student engagement in OCD, prior research has identified and categorized various types of student roles. However, although students usually change their engagement during the learning process and across learning occasions, most existing research focuses on examining unchanging student roles or developing roles in similar collaborative activities, which might overlook the probable role transitions brought by engagement changes. To fill this gap, this research proposes an integrated probabilistic clustering approach to detect student roles, role transitions, and fine-grained attributes of transitions across the asynchronous and synchronous OCD modes. The results demonstrate four roles (Knowledge Constructor, Task Follower, Isolated Explorer, and Lurker), four transition categories (Maintenance of inactive participant, Transferring to inactive participant, Maintenance of active participant, and Transferring to active participant), and the code co-occurrence structures of four transition categories. This research deepens the understanding of the complexity of student engagement in online collaborative discussions and offers both analytical and practical implications for improving student engagement.