The University of Kentucky is making a major investment in data analytics to try to improve student retention. The approach is described in an article at Inside Higher Ed:
Every time students open the app to check their course schedule or the date for the next Wildcats game, they may be faced with a quick question: Have you bought all your textbooks already? Do you own a tablet? On a scale from one to five, how stressed are you?
The university collects a student’s responses to these kinds of questions on a per-student basis. To that record, they also add a student’s interactions with the campus LMS and participation in campus events, which are tracked through a card swipe-based attendance and incentive system.
All of these systems alone represent a big investment in tracking, but analytics is about doing something with all that data. UK has made a major push to make meaning from the data by hiring a team of 15 data analysts to develop and refine a predictive model of student engagement. The end goal is to increase retention rates which, assuming they’re even marginally successful, will more than pay for the investment in all the staff and databases.
- The cost of attendance in-state is about $20,000, and $30,000 for out-of-state (source)
- The average financial aid award is about $10,000
- So net revenue per student is about $10,000-$20,000 (assuming in-state students); let’s call it $15,000 for simplicity’s sake.
- The freshman enrollment was about 4300 students
- A 1% increase in retention is 43 students
- 43 × $15,000 = $645,000 additional revenue
- $645,000 × 4 yrs = $2,580,000
- $2,580,000 ÷ 15 staff = $172,000 per additional staff line
And that’s making very conservative estimates throughout. That’s also not including the cost savings on the enrollment side of not needing to recruit as large a class.
Audrey Watters, who writes at Hack Education, has posted a transcript of a talk she gave at Columbia as part of their Conversations About Online Learning series. Setting aside the envy I have of a place that holds a lecture series about technology and higher learning, Watters goes deep on some of the implications of “data mining” in education, fleshing out some of the ways such data might be used and pointing out how risky that might be for students.
…all this data that students create, that software can track, and that engineers and educators and administrators can analyze will bring about a more “personalized,” a more responsive, a more efficient school system.
How will this magic happen? Using the same secret algorithmic sauce that companies like Google use to tailor search results and ads, and Amazon uses to sell you, well, pretty much anything. So what’s the hitch? There are at least two, according to Watters: privacy and money.
It may be obvious, but if data is going to make a big difference in student learning, that is going to require a sea change in the rules surrounding access to that data. Or is it? It appears that right now, the rules are being skirted by private companies that don’t have the same restrictions as actual schools. I suspect that most students and their parents aren’t aware of this end run around educational data privacy. It is access to this kind of data that will be necessary to assist with learning, in the absence of actual human interaction.
And the money? It’s not money in the sense of cost to students. On the contrary, most ‘big data’ education projects are free to the student, meaning someone else is paying. For now, the bills are being paid by venture capital investors that are expecting BIG returns. We’re in the early days, the thinking goes, of a major shift in the way education is done, and one of the biggest parts of this shift is the privatization of education. Sure, there has been some suggestion that these programs will lead to a system of credentials not unlike a degree, and some programs have even been rolled out. But for the most part, the schools with the biggest stakes in this territory thus far are not talking about any kind of equivalency between their live and online programs.