Skates and Stats
By Dirk L. Schaeffer
What you’ll find here.
A rather cheap little site, reporting some of the results of my
statistical analyses of figure skating judgments. This is not the world’s most exciting topic, but it isn’t
rocket science either. Have no
fear! All of the papers work on the
assumption that if it isn’t obvious, I should explain it.
Still, nothing works like feedback.
Please make whatever comments you wish in the guest book
or by e-mail at dschaeff@direct.ca.
2001A – Why do we have judges at figure skating events?
This started off as an attempt to figure out what went wrong at the 2000
World’s Men’s Long program and wound up finding the most compelling argument
in favor of raw-score systems that I’ve seen yet.
Briefly: Since the only
criteria we have for judging the quality of skaters or the ability of judges is
their agreement with each other, it follows that the scoring system that leads
to the most agreement is the one we should use.
It turns out that, although judges are trained to use ordinals, their raw
score judgments show greater
agreement.
This is a fairly long paper, about 5000
words.
Read it
here.
2000B – Tie-breakers in figure skating:
stupid, irrelevant, incompetent, biased, and counterproductive.
The title says it all. This
is a spin-off from the preceding paper, generated when it became clear that the
repeated tie-breakers involved in our present scoring systems were what was
making most of the trouble.
It’s fairly short, maybe 1400 words.
Read it here.
2000C - What can we do about dishonest judges?
The final paper in this series, built on the recognition that the
apparent ability of ordinal systems to limit the damage that can be done by
biased or dishonest judges is the only justification for their continued use
that could be made. But it too
seems more apparent than real.
Another long one, 5000 words or so
again.
Read it here.
1325
People have
viewed this site.
Entire contents of this site copyright by Dirk
L. Schaeffer
HTML rather unprofessionally coded by Anarchy
Computes!
