Location: Salon 3
This is a roundtable in the Cyberlearning 2017 Roundtable session.

Large-scale quantitative analysis of learner trajectories in informal online communities

Sayamindu Dasgupta
In this presentation, I will be describing some of the recent and ongoing quantitative studies of learner trajectories in the Scratch online community. I will present how we have been using methods and strategies from epidemiology, social policy research, etc. to understand the factors that support and influence informal learning of programming in an online community such as Scratch. As concrete examples, I plan to use two published studies — one that examines the relationship between appropriation (e.g. remixing of projects) and learning, and another that looks at the learning progress of Scratch users outside of the United States, and it’s relationship with the user-interface language that they choose to program in. The first study was published at the ACM CSCW conference, and the second study is scheduled to be presented at the ACM Learning@Scale conference. I hope that this presentation would spark discussions on possible strategies and pitfalls for doing large-scale quantitative analyses of informal learning in online communities.
Project: NSF Award #1417952 – Collaborative Research: New Pathways into Data Science: Extending the Scratch Programming Language to Enable Youth to Analyze and Visualize Their Own Learning

Exploring the Temporal Participation Patterns in the Discussion Forums of Massive Open Online Courses

Hengtao Tang
Previous research informs active forum participation as positively relating to learner achievement in Massive Open Online Courses (MOOCs) in such a way that more posts over time yield better performance, but overlooks the fact that learner participation is longitudinal and the course attritions might be gradual. Undoubtedly, understanding the temporal needs of learners becomes increasingly significant. The massive datasets generated by MOOCs make it possible to investigate the temporal behavior of learners. Therefore, this explorative research applies educational data mining techniques (i.e., longitudinal k-means cluster algorithm) to investigate the temporal patterns of learners’ forum participation in MOOCs and then identifies the influence of different longitudinal participation traits on learners’ course achievement. The findings validate the significance of focusing on temporal patterns of forum participation rather than numerical convergence of overall forum posts. This research also reveals learners persistently engaged in the discussion forum are more likely to achieve more in MOOCs. Further, the longitudinal k-means cluster algorithm is empirically promising in educational research, especially in the efforts to examine the temporal trajectory of learner activities in a course.