Posting has been extremely light lately primarily because I have been swamped with an unusually large number of competing priorities. These have included applications to doctoral programs, which in turn means that I had to take the dreaded GRE.
Preparing for the test was both tedious as a process (it has been a long, long time since I had to take a standardized test) and interesting as a social barometer of how we approach scoring and ranking people. Educational Testing Service, who administer the test, have changed their scoring methodology significantly this fall for the first time in many years, and the reasons why are clear in these comparisons of the old scores (which ranged up to 800, in 10 point increments) and the new (which range up to 170, in 1 point increments). I’ve taken these from their helpful explanatory document, which is here.
It’s clear from these figures that they had to do something. People scoring at the top end of the old quant test were effectively indistinguishable once they went above the 94th percentile, which renders the test not very useful for programs looking to select from that group. By contrast, the old verbal test was extremely sparse in terms of scores at the top end, with even a 700 landing in the 97th percentile.
These results strike me as confirmation of the fact that students are prepping much more intensively for the quant section of the test than for the verbal, which in turn has to mean that the programs they’re applying to emphasize the former over the latter. The verbal section was frustrating but ultimately quite interesting in terms of testing knowledge of how arguments are constructed, and how to draw inference from them. It seems to me we could use at least as much of that skill as straight-up calculation, but that doesn’t seem to be what admissions committees are emphasizing in the aggregate.