-----Original Message-----
From: Karl L. Wuensch

"Looks like Skategate is more about psychology (and political science?)
than about statistics.  Still, I am ignorant about the nature of the
data provided by judges -- do they provide both a RATING and a RANKING?  If
so, I might well agree that the rating would have more informational
content than the ranking, even if both were properly considered to be
ordinal data."

It just so happens that I came across this explanation of judging on the
nbcolympics website
(http://www.nbcolympics.com/x/f/frame.htm?u=/news/683689.asp). The
alleged interval data is based on the following scale: 

       0 = Not skated
       1 = Very poor
       2 = Poor
       3 = Average
       4 = Good
       5 = Excellent
       6 = Perfect and faultless

These certainly are not equal intervals and strangely enough the 3 out
of 6 is considered "average". There are more specifics about what to use
as a basis for judging technical merit and presentation on the site. The
judges rate every skater on both technical merit and presentation. Their
ranking follows directly from the rating unless there is a tie between
two skaters in which case the presentation score supercedes the
technical merit as basis for the ordinal. So, the ranking does flow
directly from the rating unless there is a tie, in which case the
presentation supercedes technical merit in determining the ranking. I
don't know what happens if both marks are equal for both skaters. I
think there is an argument to be made that the ratings have more
informational content than the rankings. I would guess (although I don't
know) that, ironically, the ordinal system may have been instituted in
the first place to make cheating by individual judges less likely since
an extreme rating would have less impact on the average overall marks
than the rating turned into a ranking.

Rick

Rick Froman
John Brown University
Siloam Springs, AR 72761
e-mail: [EMAIL PROTECTED]



---
You are currently subscribed to tips as: [EMAIL PROTECTED]
To unsubscribe send a blank email to [EMAIL PROTECTED]

Reply via email to