Monday, 22 May 2017

When learning analytics meet #MOOCs by @mebner #learninganalytics #liveblog

Today I had the pleasure of meeting up with Martin Ebner from the TUGraz, who gave a detailed and critical overview to show how Learning Analytics can support the future MOOC-learner as well as the future MOOC-lecturer.

Martin starts off talking abut iMOOX (with an explicit open license (creative commons), so you can use the content for free. EdX license for open courses was a conscious choice.
‘Making’ kreatives digitales gestalten met Kindern (, another famous one: Gratis online lernen. Which won them the Austrian National Price of Adult Education in 2015. So, to reach all adults a blended approach was used to get the learners from what they new (paper learning) to MOOC learning.

How did they implement learning analytics
iMOOX learning analytics prototype architecture: learners, MOOC  platform, learning analytics and prototyping, that was returned to the learners.
The webserver did log-files, the data was collected, and the learning analytics server visualised results. The LA server was developed at TUGraz (Khalil, yeah!!) with screenshots.

Some MOOC learning statements: what can we learn from these learning analytics
High dropout rate on MOOCs is a legend: statistics show that it is a myth (referring to Lackner, Ebner, Khalil (2015). Moocs as granular systems: design patterns to foster participants activity, elearning papers, 42 (2015), pp. 28-37.

Activity profile is shown: posting, reading, text files, … with different colors for certificate earners.
The decline of participation over the weeks. Very similar to other MOOCs. After the first 4 weeks learner activity is quite stable.

Video tracking was done, (Khalil, 2016, what MOOC Stakeholders can learn from learning analytics? In Lockee, Childress (Ed.). The video start is seen, but the video end rarely.
Learning success:what is it? In many cases learner activity is compared. But not clear link to which activity is related to learning success (not sure if I got this).
Learning analytics tells us how learning in classroom happens: social aspects lecture of Ebner was turned into a MOOC. The idea was to open the course to both the students and the public. The course lasts 10 weeks, with regular content, and the learners can choose how to go through the MOOC, and it includes a self-assessment per week. Each quiz could be taken 5 times. Cut-off was 75 % and if you did all the quizzes you got a certificate. When looking at the clusters (Khalil, Ebner, 2016, clustering ptterns of engagement in MOOCs: the use of learning analytics to reveal student categories).
University students had highest certification ratio, with high reading, low writing, high video watching and high quiz attempts. (really interesting).
Learning happens all the time and can be fun. (Lackner, Khalil, Ebner, 2016, How to foter forum discussions within MOOCs, a case study).

Then badges were introduced, compliant with mozilla backpack. Badges for quizzes and course finalisation. This was used for the first time in 2015. Learners who did take badges, went for all the badges. The drop out rate was much lower for those learners going for badges, then simply certification. (Inge: is this related to multiple motivation drivers?).

Gamification: was done in one MOOC. (Khalil, Ebner, 2016, how gamification can improve your MOOC student engagement). The gamification element got more learners active than in the first week! Never seen that graph in any other MOOCs. (interesting). But one setback with gamification, are those who are no longer active, as these are not engaged in the game.

Learning analytics constraints
Revealing personal information
Morality to view student data
Collecting and analysing data transparency
Deleting data policies
Who owns student data
Protection and copyright
Integrity, confidentiality and availability
Inaccurate analysis results

Benefits: potential for society
Knowledge society
Access to education (but bachelor degree at least are the most successful learners in MOOCs)
Lifelong learning
Cost reducing
Quality improvements

Learning analytics: drop out not that high, get them in for the first 4 weeks, the rest is stable.

Remark of Martin on a recent adult learning MOOC, which used local ad-hoc groups that could meet on the subject of MOOCs, which turned out to be very successful (similar to other MOOC groups that were set up).

Future actions: bridging MOOCs to prepare students prior to university. 

These are Martin's slides: