MOOCs: paradise for learning analytics

MOOCs, as the name suggests, attract massive numbers of participants that all engage in numerous activities, which usually are logged and available for review to the moderators of the course. What this means is that there are huge data sets with information on participant behavior. Not surprisingly, recent years have seen a rise in the number of journal articles and conference presentations dealing with the topic of learning analytics, which means data about learners is analyzed with the purpose of optimizing education. For example, by correlating student activities during a course with the grade or outcome of the course, one could  analyze which student activities predict course completion.

In this blog, I would like to take a look at the upcoming learning analytics (LA) conference in April 2016 (LAK2016, see this link:  and pick two interesting directions research is taking concerning the intersection of MOOCs and LA.

Direction 1: studying gaze behavior during MOOC videos

Many MOOCs include watching instructional videos as their primary learning activity. Although this fact in itself could be questioned (i.e., should MOOC participants engage in more active forms of leanring? See Koedinger et al., 2015) studying how learning from videos can be optimized is an interesting step forward. The conference contribution by Sharma et al. studies the concept of “with-me-ness” while participants watch videos, i.e., the extent to which a learners’ focus is aligned with that of the teacher in the video.  The authors designed a tool that provides a visual aid superimposed on the video when  the learner’s with-me-ness is under the average value, which is computed from the other students’ gaze behavior. The authors found that learning gains improved because of this tool.

Direction 2: reducing chaos in MOOC discussions

When discussion forums are used within MOOCs, it is easy to imagine that the number of discussion topics and the number of posts within them leads to an incomprehensible mass of activity. In one of the sessions during the Lak2016 conference, this issue of “untangling chaos” is discussed in three presentations – in particular, the researchers examine to what extent it is possible to engage in meaningful, collaborative activity. Hecking et al. examined whether social  and semantic roles (discussion patterns and topics on which participant posts messages) participants fulfill are related, and found this was not the case. Poquet and Dawson examined the potential of MOOC to develop social processes. Their results showed that this was indeed possible within a group of persistent learners. This session is not only interesting because of the topic, but also because of the advanced analyses that are employed, among which social network analysis and blockmodelling.

In short, learning analytics approaches seem to be a promising direction to answer questions about and optimize learning processes in MOOCs.


Hecking et al. (2016). Investigating Social and Semantic User Roles in MOOC Discussion Forums. To be presented at the LAK2016 conference, see

Koedinger et al. (2015). Learning is not a spectator sport: doing is better than watching for learning from a MOOC. Proceedings of the Second (2015) ACM Conference on Learning @ Scale, Pages 111-120. doi:10.1145/2724660.2724681

Poquet & Dawson. Untangling MOOC Learner Networks. To be presented at the LAK2016 conference, see

Sharma et al. (2016). A Gaze-based Learning Analytics Model: In-Video Visual Feedback to Improve Learner’s Attention in MOOCs. To be presented at the LAK2016 conference, see

CC BY 4.0 This work is licensed under a Creative Commons Attribution 4.0 International License.

Leave a Reply

Your email address will not be published. Required fields are marked *