Re: Division II PATH Stats

edcohn · Sat, 14 Apr 2001 12:14:53 -0700

Like a lot of people who've posted on this, I'm
somewhat skeptical of the PATH statistic. I don't think
it's a _bad_ thing, btw: it clearly has its
advantages, and I'm sufficiently stat-obsessed to enjoy
seeing another statistic. But I'm wary of treating any
individual statistic as terribly meaningful, and PATH
presents its share of difficulties.

To begin with,
as other people have mentioned, I'm not convinced
that the relationship between a player's scoring and
the scoring of his/her team-mates is quite as simple
as PATH would suggest. I'm somewhat skeptical of the
PATH statistics for my Chicago team from last weekend,
for instance. Christian Kammerer, our fourth player,
is one of the deepest players I know in the
categories he specializes in; I'm not convinced that he
would have doubled his score if he were playing alone.
Jeff Bennett and I were competing for a lot of the
same points, since we're both good at history, current
events, and geography; PATH probably underestimates the
amount by which our scoring would rise if we were
playing alone. And then, of course, there's Andrew. He
scored points in everyone else's specialties over the
course of the tournament, but his strongest category was
lit--the main weakness of our sectionals team. So he
scored a bunch of points that other people might have
gotten otherwise, and also answered a lot of questions
that would probably have gone to the other team or
gone unanswered if he hadn't been there. My guess is
that PATH did a better job of estimating how his score
would change if he were playing alone than it did in
estimating his impact on the scoring of the rest of the
team, but that's just a guess. (I won't pretend that
any team with Andrew on it was typical of the rest of
the field, btw, but I suspect that similar problems
would pop up in a lot of situations. Chicago just
happens to be the team I know best.) 

Another
PATH-related issue: it corrects a player's scoring based on
the scoring of his/her team-mates, but not on the
scoring of his/her opponents. In an event like the
Division I championship, I suspect that strength of
schedule plays as large a role as strength of team-mates
in its effects on a player's scoring. It would
therefore seem strange to me to hand out Division I
all-star awards based on PATH, even if--in practice--most
of the people given awards are the same either
way.

Granted, I realize that no tournaments release the
statistics necessary to determine the effects of opponents'
scoring on an individual player's scoring. And I realize
that giving out all-star awards by PPG also favors
players with weaker opponents, all else being equal. But
if you're going to hand out awards based on an
imperfect statistic, I'd rather use PPG--which at least
bears a closer resemblance to what actually happened
over the course of the tournament.

[Note: I'm
not saying that because we can't correct for all the
problems with a statistic we shouldn't correct for any of
them. That would be silly. What I am saying is that we
shouldn't correct for just some of the problems with a
statistic and then treat the new stat as if it were a
perfect metric for a player's individual ability. Please
note that I'm not accusing Samer of doing this,
either. But I don't think it would be a good thing if
PATH replaced PPG as the standard measure of a
player's performance, or even if it were treated as the
equal of PPG in its usefulness or
significance.]

(cont.)