re: "To the point."

[Kestutis.Kveraga@dartmouth.edu (Kestutis Kveraga)]

--- You wrote:
Actually, my knowledge of statistical theory (and it has been a while, I could
be wrong, I'm sure others on the list have a better knowledge than I)
 makes me think that this game is the worst example you could've picked. 
Mostly because it's not a typical game (i.e. Walker not doing the things he
will do in almost every other game would effect your numbers).  Still, it backs
up the fact that Kenny had a good game, and that Milt, like the rest of this
preseason, hasn't been all that effective.  It's still just one game though,
and maybe a game that's not going to be indicative of the rest of the season.
--- end of quote ---

The bigger problem with a +/- analysis is that it doesn't account for other
factors that may covary with the variable of interest (i.e., who's playing PG,
Kenny or Milt). The most obvious of these covariates is who else is playing
with the PG. If Kenny tends to play with the starters a lot, and Milt with the
scrubs, it may very well turn out that Kenny has a much better rating, even
though it may have nothing to do with Kenny. Other factors that may have a
large effect on the +/- rating is whom the player is playing *against*, as well
as the particular game situation. For example, Kedrick probably had a -25
rating in the San Antonio game, because he only played in the 4th, when the
game was over and the Spurs were piling it on. Basically,  for such an analysis
to be meaningful, it'd have to account for all these other variables. 
Kestas