Statistical error in judging

[mathman444]

I didn't mean for my language to be denigrating. It was my intent to say that the difference between the old and the interim systems, from a statistical point of view, was small and unimportant. (Oh wait, is that what "denigrating" means?).

I agree that neither of them answers well to the charge of permitting judging bias.

About the quid pro quo, my next point was that, in contrast, the difference between an ordinal based system and a points based system was <em>not</em> mere twiddling but really had content, whether for better or worse.

Never mind. The real reason that I am responding here is to ask what EDB means.

Respectfully and disagreeably yours,

Mathman (anxiously awaiting his next writing lesson).

[rgirl181]

Come now, Mathman, everybody knows Rgirl is an EDB
Just as everyone knows that the equation for this thread can be expressed as: RG<~~>MM = A2D/R
Rgirl MBEDB

[mathman444]

Let's see. The B part is obvious. Still working on the rest.

[3axel]

You can slice it and dice it anyway you want, but if the computer picks include the 9 most likely to favor skater A and leaves out the 6 most likely to favor skater B then the computer will likely skew the result, no matter how much better skater B may perform. This would make all your arguments add up to meaningless technoblab.

3axel

[mathman444]

No, no, no, 3Axel! (Rgirl, I blame <em>you</em> for this!) That is a logical fallacy. A seductive one to be sure, but a logical fallacy nontheless. Think about it.

What if the draw were done not by computer but by pulling names out of a hat?

What if we pulled names out of a hat 6 months before the event.

It would STILL be the case that the random draw might select from the pool 9 judges that were favorable to skater A, and not select those judges that were more favorable to skater B.

You are playing into Speedy's hands if you dwell on this argument and let him thereby obscure the real issue, which is secrecy.

Mathman

[engrsktr]

wow you mathematicians have made the engineer feel somewhat behind in her math training... :lol:

I think the problem isn't so much one of numbers as it is of people.... you can't possibly hope to pull a fish out of a contaminated pond (at least not one that is normal!)....
I believe a judges history is EXTREMELY important and should be monitored closely.. probably moreso than it is now....
I think the institute of judging is losing it's focus. judges are so concerned about where another judge is placing skater A vs skater B that they may end up placing skater A or B as a RESULT of the other judge (or where they think the other judge will be placing them.)
what i"m saying is that there is too much "off-setting" going on... the eastern block is always accused (and many times rightly so) of block judging and such.... but are the North American judges not also guilty if they judge ACCORDING to what they think the traditional block will do?? I can see that they want to keep it fair and offset the potential bad judging but it's not the judges responsibility....
And the biased judging isn't always obvious.... skating is a very subjective sport.... about the only thing that is really objective is whether a skater falls on an element....
who is to say who is a better artistic skater - kristi or nancy?
this is a sport judged by humans who judge human characteristics....there was a discussion about why certain athletes are more able to be indentified with by their fans.... a judge is a human and they have their favorites I'm sure.... likes and dislikes.... who is to say that the artistic mark in the pairs event was fair for M&D vs the marks for G&G??
it seems that everyone is trying to eliminate the subjectivity of a sport that can rely on very little else for scoring (according to the elements that are considered to be important).

[mathman444]

ITA, Engrsktr. What is your take on the new "points-per-element" system that the ISU wants to try out next year?

Mathman

[engrsktr]

well it sounds good at first.... but then you think - there are TWO parts to skating.... technically it might eliminate a lot of bias because there will be an actual number associated with the elements... provided the elements are given the proper credit (and how one makes sure of that....??) it should be ok.... but the problem is that there is no way to judge a skater artistically with a points system.... it's just impossible... sure you can look out for certain things but there is no REAL way to objectively judge this...
points are only viable when you can assign them.....

[engrsktr]

oops I forgot to mention....

if a system of points is to be used for one aspect, it would be inconsistent to not have an objective system all around.....
the system only works if it's complete.... the artistic mark still leaves the door open....

[DORISPULASKI]

IMO, The points per element method is very easy to fudge. If I know I am supposed to support Jeannie and to put down Flossie, I just give Jeannie +.3 on every completed element and give Flossie -.3 on every completed element.. In some ways, this is easier than the current system to fudge. If I think other judges won't properly diss Flossie, I can also be sure to give MaryJo +.2 on all completed elements, which will keep MJ better than Flossie while still being worse than Jeannie.

What could be easier?

dpp