Somehow I lost half of this post between the writing and the sending.
I wrote:
It's sounding to me as though there's a notion that the reduction in games
won (by the Yankees) is either due to regression towards the mean, or to
some other identifiable cause. But regression is merely a statistical
phenomenon, not an actual cause in itself. The fact that the number of games
won by the Yankees this year will be less than last year's extreme is surely
due to a large number of actual causes, and yet reflects the statistical
regression phenomenon. The fact that they won so many last year was surely
due to a large number of factors going their way all at once. Many of those
factors were beyond their control, and not surprisingly, didn't go that way
(at least as strongly) this year.
====================
I meant to add that this multiple causation is why we have regression
towards the mean. The fact that one can identify some of those factors
doesn't make this any less an example of regression towards the mean.
I also meant to point out that William Wozniak's question seems more
important - How do we identify which population mean is the appropriate one
to expect regression towards? Surely McGwire and Sosa have longrun mean
HR/season numbers different from those of the major leaguers in general, and
therefore there is no reason to expect their numbers in a particular season
to regress towards the ML mean. I'm not yet sure that 70/66 (for
McGwire/Sosa) amount to extreme numbers, and I have no predictions about
regression in their cases. I wouldn't be all that surprised to see either of
them break their own records from 1998.
Similarly, the longrun wins/season we'd expect from these Yankees is not 82
(a .500 season), but some larger number (perhaps closer to the 100-109
they're likely to actually achieve this year). Nonetheless, I'm confident
that the 125 games won last year was a statistical aberration - an extreme
number even for the Yankees. I put that assumption on the line this year,
and I'd be happy to do so again: I predict that they won't win 125 games
next year either.
---------------
What I won't dispute is Tim Shearon's claim about the Braves. Much as I
hate to admit it, Chipper Jones has to be the NL MVP, in light of his
heroics in picking up the load from those lost stars.
Paul Smith
Alverno College
Milwaukee (original home of the Braves)
