I didn't. I did however assume a correlation between SAT scores and ADMISSIONS rates. I made no assumption about which of those admitted graduated. It's not relevant to my point.
If 50%-70% of HS students take the test (I'm guessing here, but it seems reasonable) and the top 50-70% of those are admitted and somewhere around 50-70% of those graduate my argument that a score from the 80th percentile of SAT takers will be around the middle or lower of scores of college graduates holds.