@Euxox I think you are way off on your logic. I've always agreed that there are freak outliers in both directions (some National Merit go to state schools bcus its free, and some people get into Ivy's that end up being terrible students out of their own poor judgement rather than competition). It gets harder to argue when you start talking about large chunks of people across representative groups of schools.
But here's an example:
University of Hawaii (Manoa) is right in the middle of the USNAWR's (admittedly flawed) rankings. Their ACT 25/75th is a 21-26. So I'll be
really generous and say that a big sample of top students there would be a full deviation above their school's 75th percentile, giving them ACT scores of about 30 and placing them at about 96th percentile in their school (it's more likely the people making STEM A's there average around 28).
Compare to WUSTL, a USNAWR next-best-10 known for premeds where the 25/75th is 32-34. Only the top third or so make it through the premed track - and of the two thirds that drop, the vast majority do so after giving their all and getting C's, not due to slacking or other non-competitive reasons.
By the numbers, the odds are hugely stacked against the Top UH'ers (avg. 28ers) making it through the WUSTL curriculum (which cuts over half of avg. 33ers). Granted there will likely be a couple outliers at both schools (People with 35s at WUSTL who end up dropping and people with 28s at UH that could make it, etc) but as a group there is no way they'd survive at the same rates as many imply while defending less prestigious schools.
Again I have nothing against either set of schools, average state or big name private. I think that either school's best and brightest would make for wonderful doctors.
My major concern is just that taking only the best and brightest from the full range of schools makes no sense. A system where only the top third at Ivy and at State both go to medical schools makes less statistical sense than sending three quarters of the former and one tenth of the latter. It's been said many times before: if you really want to be a doctor, ideally go to a top school and get a 4.0. If you are realistic and want to be a doctor, it's much safer to go somewhere that you know you'll stand at the top of the class, than to put yourself among geniuses and pray you come out above average.
If my younger sibling had a 30 ACT and was choosing between local (very decent) state school or somewhere like Hopkins, and they wanted the best shot at a career in medicine, by the numbers they should go state.
This is also seen in the flat GPAs and very different MCATs across schools. Stanford (MCAT 33-41) and Loma Linda (25 - 36)
both have the same GPA range 3.5-4.0. With my view of things this makes sense: Stanford is taking mostly top-third students from top-tier universities, who bring along top percentile test scores, while Loma Linda is taking a lot of top-third students from average universities, who bring along much more average test scores. Explaining that score disparity despite equal GPAs (which is found across most near top - near bottom comparisons) becomes extremely difficult if you don't invoke
massive differences in capability between the undergrad institutions.