Will big data answer our questions about online learning?
Fain, P. (2012) Big data’s arrival, Inside Higher Education, February 1
Sorry for the one week gap in posting, but I’ve been working on a report on the future of learning management systems which was due last weekend. I’m now catching up on some of the stuff that came in last week.
This is an interesting report on a significant subject: taking data from a very large number of online courses and aggregating them to see if they can identify significant factors that influence student performance, then using the results to modify or change practice.
Under the auspices of a WCET managed project, six institutions with major online programs have aggregated their data, covering more than 600,000 online students or more than 3 million student course records.
One of the first findings is that ‘at-risk students do better if they ease into online education with a small number of courses, which flies in the face of widely-held belief in the benefits of full student immersion.’
Well, I have to say that the first part of the statement didn’t surprise me, but the second part did. Students who take online courses tend to be working part-time, tend to be older and with families, and hence will do better if they don’t take a full course load online, especially in their first year. Furthermore, students need time to adjust to online learning. Working more independently without regular face-to-face contact with an instructor takes time to get used to. Throwing such students into a full course load entirely online is asking for trouble. It doesn’t need 3 million student records to confirm a fact that has been known for some time. There is a great deal of research for instance on factors associated with non-completion or drop-out in distance education.
Unfortunately though most administrators and instructors don’t read the research, and the US system penalises students financially if they don’t take a full course load. So if ‘big data’ can emphasise something that is already well known but is not acted upon, and lead to action where previous knowledge was ignored, it may be worthwhile.
However, it does illustrate the point that ‘big data’ is only useful if it tells us not only something we didn’t know already, but answers questions that we can do something about. In other words, big data is only valuable if we ask important questions and it can answer those questions. This means collecting data in such a way that it will answer our questions. Going backwards from data collected for other purposes to questions it might answer is a dangerous practice, because you only see what you want to see.
This is not to criticize data analytics, but merely to ask that we deal with this issue, as with all other issues in online learning, thoughtfully, and not just throw large amounts of data at a problem then jump to conclusions without cross-checking it with other sources, such as research/experience from practitioners. In the meantime, let’s give big data a chance and see what it comes up with.