First off, you should head into the medical student forum and see how people feel about clerkship evaluations.
Second, you've now shifted a bit to specify "top" medical schools that are swimming in high stat applications. In that case you are right, given that high standardized test scores are essentially *the* markers of academic performance in medicine, the selecting applicants with the highest initial performance on a high-stakes standardized test is a good move. And indeed, the schools that have a mission to education the next generation of academic subspecialists, field leaders, and department chairs take this approach with gusto. In fact, I doubt you could pry their adcoms off the MCAT scores of their applicants with a crowbar.
But what if your school's mission is to educate physicians who will go on to serve a specific geographic area, like eastern North Carolina or eastern Tennessee? At that point your major concern with metrics is
good enough, as the rest of the application will be much more important to discern mission fit.
I do note that in Figure 14 of the
AAMC report you linked earlier that 75% of students with MCATs below 492 receive a passing score on Step 2 CK. That's a MCAT range most schools would not go near. And once you get above 499-500 most everyone passes. That doesn't stop the average matriculant MCAT from being much higher.
When I sat on an admissions committee for the first time I had a great respect for the predictive power of metrics. I thought I could use numbers to divine the fate of each person who joined the class. In retrospect that was a pretty naive take. Look back at Figure 14 for a MCAT of 511. The 75th percentile Step 2 score is 257, the 25th percentile is 239. So for the same MCAT about a quarter of the students will have a Step score that puts them in contention for highly competitive residencies, and about a quarter will be hoping for a university IM spot in the upper midwest.
Point is, while there is a positive correlation between metrics and academic performance, the modest slope and high amount of variability within each score make these tools challenging to use when you're operating as most admissions committees do: evaluating one applicant at a time.
Perhaps I have missed it, but I don't recall seeing the argument that metrics are "worthless." Over the years I have heard cogent arguments that metrics do not define one's potential, and that there is a point of diminishing return for them that is well below the average numbers for matriculants. I think both are reasonable assertions.
