> 4. The issue with most “research” that claims bias in elite school admissions is that they normalize the data based on grades and scores. The scores and grades at elite schools are heavily condensed at the high end, so they aren’t really the main differentiators (unless you are a recruited athlete). The stuff that the researchers can’t easily quantify often swings the decision between reject and admit.

How do we square scores being heavily condensed at the high end with still seeing large between-group score differences? We'd expect, surely, to see smaller between-group score differences if one end is cut off.