The physics GRE...what a can of worms. At some point I found a number of references supporting the view that the physics GRE is, at best, weakly positively correlated with graduate school success, but I'm not feeling motivated to find them again. Anecdotally, I know a number of people who scored well (>800, 73rd percentile) on the physics GRE and successfully completed a PhD. I also know a number of people who scored extremely dismally on the physics GRE (<650, below the 40th percentile) and also successfully completed a PhD.
Were there significant correlations in the quality of the PhD? Let's say the number of people whose scores I know is 8. 2 very high, 1 medium, 5 dismal. (Note: These people are PhD students at top 25 universities. People with very low physics GRE scores are still accepted to some very top tier programs.) Of the 3 people who completed their PhD's fastest (in our class of 35), 1 was a high scorer, 2 were dismal scorers. One of the low scorers had very high impact publications and was highly encouraged to pursue the academic route. The other low scorer is also pursuing an academic route, currently in the post doc phase. The high scorer attempted an academic post doc and quit within 6 months to go into industry. The other high scorer, the medium scorer and 2 of the other low scorers completed their PhD's. The medium scorer went into industry. The high and 2 low scorers went the academic route. I'll break it down into more of a table form.
High scorer 1: Fast, medium impact PhD. Academic postdoc, quit for industry in 6 mo.
High scorer 2: Longish PhD. Academic postdoc
Medium scorer: Longish PhD. Industry
Low scorer 1: Fast, high impact PhD. Encouraged to go to academics, went to industry
Low scorer 2: Fast PhD. Academic post doc
Low scorer 3: Medium length PhD. Academic post doc
Low scorer 4: Longish, medium impact PhD. Academic post doc at Harvard
Low scorer 5: Still finishing PhD
(Note, I don't have information about the "impact" of most of these PhDs.)
Conclusions from that data? Umm, too low N. But certainly not a perfect correlation between high scores and high impact, quick PhD's followed by high impact academic careers. And the most encouraging thing from the data is to see that excellent institutions recognize the lack of correlation enough to still accept people with really terrible scores.
I've also read some statistics about the physics GRE, but again, I apologize for not looking up the references again. Women score an average of *100 points* lower than men. And international students make up an extremely disproportionate section of the top 50% (disproportionate to numbers taking the exam). Put another way, US students make up a disproportionate section of the lower 50th percentile.
So, what does the physics GRE indicate? Stay tuned...