Re: Memo to the College of Arrogance

Keith Willoughby wrote:
Kenny1111 <kcyanks1@xxxxxxxxxxx> writes:

But by pointing out how insignificant his post-season sample size is,
I would hope that at least one person reading my rantings might
realize that sample size/randomness are huge issues.

And you're correct to do so, and if you don't mind having (or ignoring)
the "athletes are human" argument each time, then that's fine. On the
other hand, choosing not to define your argument in narrow statistical
validity terms, and instead incorporating the human angle, might
actually convince more people :)


Right. I'm not trying to be completely dismissive of the human element.
I'm trying to temper the felt need to find human-element justifications
for observed variations, where there is a significant probability that
the variations are due to randomness. I might be getting myself into
more trouble by doing this, but suppose you have p=.8 (probability
variations that far away from the mean will occur do to randomness)
along with some "human" factors that make you believe that there
actually is statistically significant difference from the mean. I would
be very inclined to not put much weight into those human factors, since
it's so likely that the outcomes could be random. Then suppose you have
p=.1. While that's borderline for what one might consider statistically
significant, one might be more inclined to think the "other factors"
that we cannot measure might be playing a roll. I'm just thinking out
loud here, so please let me know if I'm know saying things that would be
statistically false.

I have a lot of sympathy with the people who complain about treating
players as numbers (more than I used to, certainly). I think it starts
when people arguing in favour of the statistical angle move from the
general to the specific - from a group of players to a single
player. "In general, because of the inherent sample-size issue of
separating post-season numbers from career numbers, you'd be better off
assuming a good player simply had bad luck in the post-season" is always
going to be more palatable than "Tom Gordon probably had bad luck in the
post-season", because Tom Gordon is a human being in a way that "the
typical player" isn't.

Well that's always a problem when using statistics. You can always say
"but this case is different," and come up with an anecdote about why
that case is different that seems convincing. I think baseball fans
take this route way too easily, and almost always search for
non-statistical explanations for statistical variations. Gordon's
throwing up is a very appealing non-statistical justification for
something we are observing. At the same time, the variation in his
post-season and regular-season stats is probably so clearly within the
normal range of variation (I don't know the exact probability we'd
observe the variation he displayed), that to me, to give credit to the
external explanation over the statistical explanation in this case would
be giving the external explanation too much weight. (More like the p=.8
case than the p=.1 case, to use my example above--I am *not* saying
though that I've calculated it and determined this to be true.)
.

Relevant Pages

  • Re: For us mature fanz, (and dont call us dinasours)...
    ... to someone that bats deep in the lineup. ... Compare the leadoff player with the same OBA with a higher versus lower slugging percentage. ... While there generalized statistics are interesting, they provide less details than looking at individualized stats, and then... ...
    (alt.sports.baseball.ny-mets)
  • Re: Frequency of different bad beats with other statistical poker anomalies...
    ... player who is getting 10% less pocket pairs than the statistical norm, ... all other statistics for this player were within 3% of the norm that ... the game wouldn't really even be about skill at all. ...
    (rec.gambling.poker)
  • Re: The Jeopardy Answer is??...
    ... >>> bail out early and you've been saying Bernie was toast ever since I've ... >>> discern a pattern that Jeter was never as good an offensive player as ... >> statistics don't tell the whole story, but, to the extent that they do ... > just bashing me with the usual rotonerd nonsense. ...
    (alt.sports.baseball.ny-yankees)
  • Re: Just Arod being Arod
    ... Heart, guts, nerve. ... they will lead to better or worse statistics. ... Player B has very little heart, guts, and nerve. ... Go read some of the posts about Cairo. ...
    (alt.sports.baseball.ny-yankees)
  • Re: Just Arod being Arod
    ... they will lead to better or worse statistics. ... Player B has very little heart, guts, and nerve. ... ARod has a hell of a long way to go to match Reggie in Yankee lore ... pitches, it's in his stats. ...
    (alt.sports.baseball.ny-yankees)