Learning analytics and educational research – what’s new?
A brief note that I thought I’d post to see what people think.
[1] A research team conducts an investigation into some aspect of the effectiveness of teaching and learning. It’s really not important what the details are. They analyse their data, using some techniques whose details don’t matter, find some patterns they consider to be significant, which they are able to report to fellow researchers.
Is this a “learning analytics system”? If so, why? If not, why not?
OK, try this:
[2] The data is textual, audio and video, analysed using qualitative data analysis, which as trained social scientists they can do with a high degree of rigour.
Is this a “learning analytics system”? If so, why? If not, why not?
OK, try this:
[3] A machine learning team demonstrates that with their expertise in data curation and AI, this human coding can be automated with 85% accuracy.
Is this a “learning analytics system”? If so, why? If not, why not?
[4] The data is analysed fully automatically, with no hand-curation, and made instantly available to the researchers.
Is this a “learning analytics system”? If so, why? If not, why not?
[5] The data is analysed fully automatically, and made available to the educators and/or students involved in the context.
Is this a “learning analytics system”? If so, why? If not, why not?
[6] The data is analysed fully automatically, and made available to the educators and/or students, who can demonstrate that they can make an appropriate interpretation and intervention.
Is this a “learning analytics system”? If so, why? If not, why not?
In my view, we only get to a functioning “learning analytics system” when we hit [5], and then we hope to get to actionable insight in [6]. We are in learning analytics research in 3-4. Before then, it’s educational and learning sciences research — vital for clarifying the complexities of the challenge for future learning analytics systems researchers and developers, especially if the results are communicated in a form that learning analytics researchers and developers can engage with, part of the transdisciplinary dialogue central to the field.
As in any kind of intelligence analysis, there will always be a vital role for ‘power analysts’ who can wield powerful tools to examine the data from multiple angles and levels of detail. But the promise of Learning Analytics (for me) is that insights are made available to the stakeholders who constitute the learning context, because the analytic capability (grounded in good research) is embedded in the platform and delivered in accessible, actionable form.
Is there a risk that if we start to call 1—2 learning analytics, that we are emptying the concept of some important meaning, which breeds scepticism that this is anything more than a rebranding?
Leave a Reply