‘it’s like rating a blender’
Org theorists know a thing or two about what happens when you rate things. People change their behavior. In this case, that’s the point — Arne Duncan et al. are hoping that the ratings will create incentives for colleges to graduate more students with less debt and higher post-graduation incomes.
Now, those are obviously not objectionable goals. There are some clear challenges in adjusting for the expected performance of different student bodies, and worries about disincentives to go into low-paying fields like teaching or social work, but who doesn’t want college to be more affordable, somehow?*
The big problem is the outcome that is missing in there: students who have learned things. If you create a system that measures access, completion, debt, and eventual income, and it has any teeth at all, you will get colleges that aim for those things. Unfortunately, those things have a limited relationship to actual learning. Where one conflicts with the other, learning will lose.
Of course, I’m kind of hesitant to say that, because heaven knows what would happen if we started trying to measure learning outcomes at the federal level. No Young Adult Left Behind, I guess. Coursera can sell us the curriculum.
* Another problem worth mentioning is that many adults without degrees don’t see graduation rates and average student debt levels as relevant to their college decision — they think it depends on them, not the school.