Why do Learning Analytics research?
(Invited remarks at the launch of another university’s Learning Analytics research network)
Many universities base their learning analytics capability in a teaching and learning service centre, or in an IT business intelligence unit. You have already got some analytics running as a service, but you have now decided to launch a learning analytics research network.
So, why would you want to establish a research program around learning analytics?
What kinds of research are you talking about anyway?
There are many reasons to conduct learning analytics research, not all of which may apply to you, but I’ll leave you to figure that out. So, how do these proposals sit with you?
You want to invent new kinds of analytics that have yet to be productized. These might be variations on the data or dashboards your purchased products provide, or they might be completely new concepts.
Why might you want to go beyond out-of-the-box solutions?
- You don’t want a black box proprietary ‘solution’, but one you can fully understand, and modify. Perhaps you want to go open source as well, and initiate or join a dynamic developer community?
- The analytics have yet to be invented. This might be simply because you’re first person who wants to make sense of activity traces. For instance, you might use a simulation tool that can also export rich data, but nobody’s analysed it.
- Or, you may be interested in a form of learning that isn’t mainstream (so no products), or particularly complex. For example, you might want to understand the potential of automated analysis of student writing or feedback; or you value direct, embodied learning activity, not just traces from mediated interaction online. The learning analytics conference and journal offer myriad examples of how new kinds of sensors in the physical world make learning digitally visible and analysable in completely new ways.
Learning Analytics is inherently multidisciplinary at the intersection of (minimally) education + computing (but today this expands into many other disciplines including design thinking, ethics, organisational strategy, user experience). This requires people who have depth of knowledge in at least one discipline, and who relish working across disciplinary boundaries. So you may fit into this box as a researcher, though researchers are not the only people who can do this of course, and in fact many academics are guilty of not wanting to make themselves vulnerable by stepping out of their intellectual comfort zones. A university should actively promote and incentivise such boundary-crossing if it is to stay relevant to real world problems – because surprise surprise, the real world isn’t segmented into neat disciplines, but tangles them.
You want to make evidence-based claims about what works, and fails, and why. If you want to argue that learning analytics made a difference, how well will that evidence withstand scrutiny? Researchers know how to design interventions in complex social contexts, and assess their impact.
Moreover, you no doubt want to publish your work, which requires you to use theories, methodologies and tools with a rigour that meets the standards of peer reviewed publication.
Furthermore, you must meet the standards of your Human Research Ethics board. Again, academic researchers will know how to prepare and defend those submissions.
Your learning analytics research might take many forms, depending on how you answer questions like these…
- You want to see education transition into a “Fourth Paradigm” data-intensive field of scientific inquiry, with power tools for researchers to study learning activity in high definition. Is theory now dead — the data speaks for itself? Or do you need theory to give your data value and meaning?
- Who is the user? Are your analytics solely to illuminate student activity for an educator, or researcher? What are the implications of designing automated feedback to students based on analytics? If all you do is provide power tools to the researchers, is it really learning analytics? They’ve always studied learners and educators.
- You are excited by a particular theory, pedagogy, learning design approach, or assessment strategy — but nobody has yet invented learning analytics for it. Break new ground by analysing the requirements that such an approach might demand of analytics, the challenges of sourcing or designing data, and of detecting salient patterns in it.
- Are you interested in mining ‘data exhaust’ (the activity trace by-products of student activity), or are you designing your data from scratch?
- How hard is it to explain how your analytics work, and to whom (educator? student?), and does it matter if they understand? Who is accountable if it malfunctions?
- Are you ready to start supervising learning analytics PhDs?
- Do you want to work quietly in a lab with small scale prototypes, or deploy tools with real academics and paying students? How will you gather data? Do experiments with conditions? What will count as evidence of impact?
- Which kinds of learning analytics will you specialise in? You can’t be great at everything, so play to your institutional strengths and focus.
- Will the university provide starter grants of any sort to get you going?
- What’s the relationship between academic researchers and your service units who work with (and may be gatekeepers to) student data?
I hope they’re good questions to start chewing on!
Leave a Reply