I understand the excitement about #GRExit, and it makes for good departmental PR, but where is the evidence that it is not a useful predictor of grad school success? 1/2

Many are making the mistake of looking at students accepted into programs and saying their GRE scores don't predict performance. But that's not the issue! 2/

Rather, you want to know how otherwise similar students with lower GRE scores who *didn't* get accepted would have fared if admitted. This is an unobservable counterfactual. 3/

It's the same fallacy that leads Google to falsely conclude that GPA is irrelevant to job performance - https://www.nytimes.com/2013/06/20/business/in-head-hunting-big-data-may-not-be-such-a-big-deal.html?partner=socialflow&smid=tw-nytimesbusiness. Sure, conditional on being hired at Google, it may not matter, but that is a highly select group. 4/

One could perhaps make a reasonable argument that GRE scores are highly correlated with GPA, and hence aren't worth the additional information they provide. But then we are back to the problem of having to make bias-prone judgments about strength of school, etc. 5/

If we are trying to diversify our graduate student pool, getting *less* information seems like the wrong direction. 6/

Yes, cost is an issue. But perhaps then departments should put their money where their mouth is and subsidize the cost for applicants instead of ditching it? 7/

And boy, will I have egg on my face if/when @publichealthumn decides to go #GRExit. 8/