Re: Division II PATH Stats

matt_reece · Fri, 13 Apr 2001 22:13:02 -0700

Hello all,

This is the first message I
have posted to this list. The first thing I noticed
was that Yahoo seems to lack a good quoting
mechanism, and this text box is too small; subsequently, my
browser crashed and X died, so now I'm typing this
message for the second time. Forgive me if I sound a bit
irritated.

I'm writing because I don't like this "PATH"
statistic very much. And it isn't just because I dropped a
bit in the rankings when PATH was used. Really, it
isn't. ;-) Or so I tell myself.

My objection is
essentially that raised already. If I have teammates who
answer pop culture, economics, or poltics questions, it
makes little difference in my average points, because I
would have been unlikely to answer them anyway. There
are some things, like physics, that I am slightly
more likely to be able to answer questions about, and
here I would have a real difference in average points
per game if a teammate consistently answered
them.

I am not a statistician, but I don't think the "law
of large numbers" argument given earlier is valid.
To abuse some terminology, the issue here is a
systematic bias, not a statistical one. The correlation
between team member's knowledge is important. I'm too
tired at the moment to try to make this much more
precise, and in general too apathetic, but here's an
example:

Consider two people. The first has a team member who is
good at the same subjects, but answers few questions.
The second has team members who are good at very
different subjects, and answer many questions. PATH will
adjust the second player's score more, although the
first is likely to have a larger
effect.

("Consider two people"?? I sound too much like a
mathematician. If I begin mumbling "to first order" or "by the
axiom of choice," someone remind me I'm supposed to be
talking about the real world.)

In short (I had
much more here, before I lost my original message, but
I don't feel like going through this again) any
true indication of an adjusted points-per-game needs
some measure of correlation between one's area of
expertise and those of teammates, which is much trickier.
As it is, I find it hard to place much stock in
PATH; I would agree that there could be some deficiency
in points per game, but overcoming it requires a
more sophisticated solution.

I don't care too
much for individual stats anyway. I've always enjoyed
team competitions far more than individual ones, and
so it does seem nice in a way to try to adjust
statistics for teamwork. But the true indications of good
teamwork on a team - knowing what sort of questions one's
fellow players are likely to answer and deferring to
them when they have more knowledge, listening to
everyone's input on bonuses - are difficult to measure, and
naively giving everyone points for anything their
teammates answer is not a good alternative.

There's
my $0.02 or so... 

Matt Reece

(i.e., a
random Chicago first-year most of you have never heard
of)

<<I have complete Div. II stats for PATH, using the
stuff from naqt.com. Here
 is the Top Ten (of
97):

 1. Matt Weiner, Pitt, 110.53
 2. Jacob
Mikanowski, Prin, 91.95
 3. Wesley Mathews, Oxy, 65.72

4. Justin Sausville, South, 61.12
 5. Tim Youker,
Stan, 59.74
 6. Matt Reece, Chi, 59.63
 7. Vinita
Kailasanath, Stan, 58.40
 8. Amar Hatti, Cal, 55.56
 9.
Mike Davies, Pitt, 50.49
 10. Paul Nelson, Rolla,
48.78

 Average PATH was 26.30

 Andy >>