Learning analytics and #MOOCs #Futurelearn

Dragan Gašević talked about Learning analytics and MOOCs: what we have learned so far as the next talk at the Futurelearn Academic Network meeting in Edinburgh livestreaming at http://t.co/1lkBk6SUh4. There was a lot in his talk, and this is just what I captured liveblogging.
He identified the challenge of giving personalised feedback to students and academics in massive courses. He noted that learners are not just using the main MOOC platform but their own environments (search, blogs etc.) and it is desirable to "harness the digital footprint when they are using these digital technologies". He felt that so far learning analytics (LA) has been focused on issues of attrition and performance.
He cited an article which had analysed mentions of MOOCs in the news (using the Factiva database) and identified that overall the trend is downwards, but some specific MOOC topics have grown. He also mentioned the recent article analysing various aspects of bids for MOOC research (e.g. topic, research approach) (the article is in this issue http://www.irrodl.org/index.php/irrodl/article/viewFile/2040/3120) and went on to discuss further the types of research going on. He felt that a good deal was possibly superficial, looking at hits, page views etc. which are not cntextualised. I got quite overexcited when he cited Tom Wilson's article on information behaviour models. So I think he said that the IB field emphasised that you needed to look at context (and in my excitement stopped listening properly), and then went on to recommend using a model of self-regulated learning (and cited this article http://link.springer.com/article/10.1007%2Fs11528-014-0822-x) which meant using "products of learning" (e.g. critical reflections in blogs and comments). He felt we also needed to understand how students where judging their own learning, what standards they used.
Research needed to be carried out into the process nature or learning e.g. looking at sequences of actions, and also to attend to life-long and ubiquitous learner profiles. He felt that "visualizations of learner data can be harmful" to students e.g. if the students are looking at their position within the class (they may be misleading, demotivating etc.).
He stressed that "learning activities do not happen on a single platform", so there is a challenge collecting and aggregating data from various platforms. There are privacy and ethical issues, and the practical issues of accessing the data. He felt that there needed to be "more robust methods and data sharing approaches" and that many institutions did not have policies for sharing MOOC data (unless you do, obviously you can't really track learners through life).