Covering the Front and Back Pages of the Newspaper
November 25, 1998
BASEBALL: A statistical fallacy in the Continuing NL MVP Debate and elsewhere
From an email I wrote to Rob Neyer in 1998:
In your columns lauding McGwire over Sosa, you rely heavily on productivity-related statistics: Slg and OBP. You thus give a team of McGwires an edge over a team of Sosas, per plate appearance.
The problem is this: Sammy Sosa had about 45 more plate appearances than Mark McGwire because he was younger and healthier. A lineup of nine Sammy Sosas would require over 400 fewer at bats by its bench than a lineup of nine Big Macs -- almost a full-time players' worth. And those bench players will naturally not be comparable players to Sosa and McGwire. In real life, this means the Cards were giving extra at bats to John Mabry that the Cubs were giving to Sosa instead of, say, Lance Johnson or Brant Brown (OK, the Cubbies had better alternatives). How can this be irrelevant to Sosa's value? Granted, in this case the formula credits McGwire with more total RCs anyway, but considering this factor does narrow the productivity difference.
For more sophisticated analysis, perhaps an adjusted RC/27 measure could be devised (adding accuracy at the expense of simplicity) by averaging in a replacement-level player's productivity at the same position up to the maximum available at bats (though I recognize this would be complex -- McGwire already had over 670 plate appearances, so what's the limit?).
Anyway, thought I'd pass the idea along because the tendency to equate percentage productivity with overall production is all too easy to slip into and should be avoided.
PS -- Why does the Runs Created formula discount intentional walks? They may not be "earned" by the hitter but they still put runs on the board. Barry Bonds' intentional passes are just as much a part of his offense as bases he steals uncontested, double plays he doesn't hit into because the infield is in, walks because he was pitched around, hits resulting from a shift, etc.