The Internet, Policy & Politics Conferences

Oxford Internet Institute, University of Oxford

Rebecca Eynon, Isis Hjorth, Taha Yasseri, Nabeel Gillani: How can social interaction support learning in a MOOC?

Rebecca Eynon, Isis Hjorth, Taha Yasseri; Oxford Internet Institute

Nabeel Gillani,


There has been no shortage of polarised media hype around Massively Open Online Courses (MOOCs) - and yet, grounded research on their pedagogical effectiveness and potential is in its infancy. 

MOOCs are unique in the way that they offer an opportunity for tens of thousands of learners from diverse geographical locations with varied experience to participate and collaborate with each other. Understanding interactions within the discussion forums is an important part of understanding the learning process within such courses. Indeed, from the significant amount of research in online learning, there are already a range of social constructivist and social-cultural perspectives that can be utilised as a starting point (e.g. Goodman and Dabbish, 2011; Lave & Wenger, 1991; Siemens & Conole, 2011; Stahl et al., 2006). 

To date, research on MOOCs has rarely explored the interaction in discussion forums. This is somewhat surprising, given that discussion forums are perhaps the greatest difference between MOOCs and previous attempts at online open education. Instead, early studies have focused on learner demographics, revealing that these courses are taken primarily by those already with college degrees (Ezekiel , 2013). Others have traced the behaviour of MOOC participants through a course's lifespan by leveraging granular ``clickstream'' data (Breslow et al., 2013; Kizilcec et al., 2013) and some have started to cluster learners based on their patterns of engagement in order to predict dropouts (Yang et al., 2013). This paper aims to complement and develop this emerging area of work, by focusing on the interaction in forums. 
While much can be learned from existing research about online learning, there are three characteristics of MOOCs that present important methodological and conceptual challenges. They are:

The scale of the activity: MOOCs are, by definition, courses that tens of thousands of learners can participate in simultaneously. Yet, the scale of the interaction that results in the forums presents some methodological challenges. Typically in past research, qualitative forms of content analysis have been used to analyze interaction and learning as the size of such forums were relatively small; although a few recent studies have begun to leverage social network analysis to explore, and model, learner interactions in online settings (Haythornthwaite, 1996; Vaquero and Cebrian, 2013). Yet, for the results of such an approach to be meaningful, we need to determine what constitutes a “significant” interaction. In studies focused on learning, this is a very complex problem. In the large, crowd-like, MOOC forums, many interactions on the forum may have very little to do with learning. Indeed, previous work as shown that much forum discussion, particularly early in the course, is small-talk or chatter that is unrelated to areas of core content (Brinton, 2013). However, while not directly related to the topic, some of these interactions may form an important basis for learning to occur. 

The non-formal nature of the setting: MOOCs encompass and interesting blend of formal and informal kinds of learning activity, perhaps best defined as “non-formal” learning (Colley, 2002). That is, such courses have a number of characteristics of formal learning, but due to the lack of formal certification and open nature of the course, learners have the opportunity to focus on different aspects of the course to varying degrees, with far more freedom than is typically available to those online learner taking courses for credit. This makes it harder to determine / characterise meaningful learner engagement throughout the course, e.g. what topics do students engage in and why, and what success “looks like”, where completion is not always the main goal of initial participating.

Learner characteristics: Relatedly, the characteristics, motivations and expectations for people taking the course are likely to be far more varied in MOOCs than in formal online courses. If, and how, learners interact in the forums is likely to differ in part due to these differences. For example, some may use the forums as a place for support, others will see helping and supporting others as part of the learning process, others will be frustrated as they seem to have “missed” the invisible rules that forum discussions encompass, others will want to connect with people who are similar to them in some way (e.g. level of expertise, first language, geographical location etc) and there may be significant cultural differences and expectations on how these dialogues should take place. These differing roles are an important piece of the puzzle in understanding interaction and learning in such settings – yet little work has explored how to examine this in-depth.

In the paper, using data from one purposively selected Coursera-based MOOC that aimed to promote interaction and learning around real world business problems, we highlight some of the methodological challenges that these three characteristics of MOOCs present. We demonstrate how an iterative mixed method approach that utilizes both qualitative analysis strategies (e.g. qualitative coding of forum posts and interviews) together with quantitative approaches to analyse forum engagement from physics and engineering (e.g. models of contagion to examine information flow) provides a valuable way of understanding the different ways that people may be learning in the crowd-like setting of MOOCs. 

In using such an approach, we argue that with further research it may be possible to develop a set of quantitative metrics that can be used by researchers that enables better predictive and real-time analysis of learners’ needs and goals. With this knowledge in hand, online platforms, educational service providers, and educators will be better-equipped to both design curricula and build software systems that promote meaningful interaction and learning in these spaces. 


Breslow, L. et al., (2013) Studying Learning in the Worldwide Classroom: Research into edX's First MOOC. Research & Practice in Assessment 8, 13-25.
Ezekiel, E. J., (2013) Online education: MOOCs taken by educated few. Nature, 503, 342. 

Haythornthwaite, C. (1996). Social Network Analysis: An Approach and Technique for the Study of Information Exchange. Library and Information Science Research, 18(4), 323-342.

Kizilcec, R. F., Piece, C. & Schneider, E., (2013) Deconstructing Disengagement: Analyzing Learner Subpopulations in Massive Open Online Courses, presented at LAK '13, Leuven, Belgium, 

Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge: Cambridge University Press.

Yang, D., Sinha, T., Adamson, D. & Rose, C. P. (2013), "Turn on, Tune in, Drop out": Anticipating student dropouts in Massive Open Online Courses, presented at NIPS Workshop on Data Driven Education, Tahoe, NV.

Vaquero, L., & Cebrian, M. (2013). The rich club phenomenon in the classroom. Nature: Scientific Reports, 1-8.


Rebecca Eynon, Isis Hjorth, Taha Yasseri, Nabeel Gillani