Nebelhorn's Numerous Numbers

Dr. Dirk L. Schaeffer takes an analytical approach using statistics and scores from Nebelhorn's introduction of the new scoring system - the Code of Points (CoP).

Have a look and then feel free to leave feedback here.

OK, I read it -- and tried to understand it! I'll just throw out a general impression, and hopefully all you more knowledgeable folks can run with the ball!

If I've read this correctly, it seems that one of Schaeffer's main concerns is that CoP will have the effect of devaluing the Presentation side of the Technical/Presentation judging equation, and he offers some data to support this. Most startling, it seems that the proportion of raw marks derived from "presentation" dipped well below those derived from "technical" even in the Long Program, negating the "spirit" of the 6.0 system.

I have no comment on what the implications are for the sport if this proves over time to be true, but I do have some sense of why the presentation scores are perhaps lower than expected.

While there were always objective criteria behind the Presentation Mark, it seems to me that under the 6.0 system there was also, for better or worse, an allowance for the subjective as well. For the "whole is greater than the sum of it's parts" effect. By dividing the Presentation Mark into the 5 components of Skating Skills, Transitions, Interpretation, Choreography and Performance/Execution, each exhaustively detailed, the overall Presentation Mark has become far more objectively based. We all know that certain top skaters, even on their worst nights, never received presentation marks lower than 5.6 or so, and for a great skate were practically guaranteed 5.8s and 5.9s. On a 10 point scale, this would translate to a range of maybe 9.3 - 9.9 (or 10!) per component, and I frankly don't see that happening. Take Michelle -- even if she were to receive high marks for Skating Skills, Interpretation and Performance/Execution, it is unlikely she would do as well (based on past performances) for Transitions and Choreography. By objectifying the criteria for awarding these marks, judges are precluded from bestowing the highest possible mark based on the emotional "wow" factor -- e.g. "She made me cry with that performance, and I haven't cried for any performance since (x) won (yz); I gave that performance a 6.0 so I'm giving this one a 6.0 too!" The "wow" factor could still come into play in the awarding of points for Interpretation and Performance/Execution, but it would be far less likely to elevate the combined Presentation Total.

The good part of this, IMO, is that reputation should begin to count for less. It always frustrated me no end seeing skaters getting inflated presentation marks for impersonating a Zamboni, and looking miserable in the process! But I also do feel that there is a certain magic that can happen on the ice that sometimes transcends it's purely objective elements.

It's too late and I've only read Schaeffer's article once so I won't comment except to say thanks, Paula, for providing a forum for such a paper and posting it. I'm really looking forward to reading people's opinions on this. Just in Gisel's post there was a point I hadn't thought of before in terms of judging presentation. I hope lots of people contribute their responses and aren't put off by thinking the article will involve difficult statistics. It really doesn't and I think any part that seems dense upon first reading (sometimes I think it's the writer's style) will become apparent upon a second. Anyway, I'll be watching figure skating tomorrow so gotta get my rest! Interesting things to consider in this article.
Rgirl

Like Geseledepkat, I too found most interesting the part about the relative importance of the technical and presentation scores in the long and short programs.

The question that Dr. Schaeffer poses in that section is this: suppose that you have to predict the outcome of the competition, but you only get to see one set of marks, either the technical or the presentation, but not both. Which would you choose?

Traditionally you would want to know the technical scores for the short program and you would want to know the presentation scores for the long.

The statistics of past events (using 2001 Worlds as a typical representative) bear this out. For the short program, the correlation between the technical scores and the overall placements is about twice as strong as the correlation between the presentation scores and the overall placements. For the long program, these effects are reversed. This satisfies our intent that the short program should be the "technical" program and the long program should be the "free" program.

At Nebelhorn, however, under the CoP no such tendency was found. Overall, the technical and presentation numbers were equally effective as predictors of the final ordinals, so the distinction between the short program and the long program is lost.

Mathman