<p>
None of this matters if the sample is big enough. We just need enough samples of SAT/MCAT scores (imagine 10,000 scores from each of 100 colleges) and compare the normalized spreads (MCAT minus SAT). Same for SAT/GRE and SAT/LSAT. Thus we can get some objective numerical idea of what the schools did to the students during attendance, on average. It’s always an apple-to-apples comparison: What did the various undergrad schools do for med school applicants? For law school applicants? For those who took the GRE? That’s all we learn, but we learn something.</p>