Being both a baseball fan and medical student, I find the negative result--based on 12 players, with 10,000 total ABs among them, using compounded stats like OBP--to be highly amusing but not really informative. Everyone should know a negative result is not proof of null hypothesis, in sabermetrics or good biomedical science. I would treat it more seriously if the docs actually cites someone other than Michael Lewis for statistical analysis of baseball performance.

Well, we know walks and strikeouts are a lot more consistent from year to year than batting average and even OBP. So KK/AB or BB/AB might have been better measures of improvement.

The commentor on that site makes my first objection, which is that those looked like largely older-than-average players, who were likely declining anyways.

And, of course, it didn't include Brian Roberts 2k5, who could elevate those guys to statistical significance as a group all by himself, for all we know. Looking at Roberts, he might have earned himself another $10 million over the course of his career by being able to explain his ridiculous improvement this year on his laser surgery and/or contacts.