> As a more concrete example... $IVY_LEAGUE doesn't gain that much by having its average standard test score go from the 99th to the 99.5th percentile.
Assuming accurate tests and a normal distribution of scores, shouldn't that be approximately equivalent to going from the 80th to 90th percentile?
Given you assumptions, the difference between the 80th and 90th percentile is about twice as big as the difference beteen the 99th percentile and the 99.5th percentile.
Assuming accurate tests and a normal distribution of scores, shouldn't that be approximately equivalent to going from the 80th to 90th percentile?