CS Weekly: Bernstein: Survivorship Bias

Against the Gods: The Remarkable Story of Al Capone, Roger Clemens and Steroids.

It’s the players who hit 762 home runs or win 354 games that generate the suspicion, especially when they not only fall outside of the normal statistics, but when they can play better, and for far longer than most, to achieve such things. It’s not just that Barry Bonds hit home runs, it’s that he hit 73 (when he’d never hit more than 47) at an age when his numbers, while still good, should have been traveling in the opposite direction. It’s not just that Clemens was pitching into his 40’s, it was that he excelled, and pitched even far better than he did at times 10 years earlier.

We can fight ourselves over the logic of numbers, and even over the logic of age … but both?

Clemens argument also brings out another point.

It is difficult to conclude anything about the use of performance enhancing drugs by looking at statistics. The Rocket is correct that players go through ups and downs in their careers—the man is a historian, remember—because they always have. And it is foolish to conclude that any “up” period must be artificial.

Roger Maris never hit 40 home runs in a season, except for the one year he hit 61. Must have cheated that year, right? (Or the fact that he was in his greatest groove coincided with the fact that he was hitting in front of a hot Mickey Mantle. He was never intentionally walked.)