Irina Slutskaya and Olympics 2002

[Mathman]

It was absolutely connected with statistics, i don't understand how is not. If only average scores were implemented, without taking out maximum and minimum, Chinese team would win the Olympics.

I am just not seeing it.

I went over the protocols for the 2018 Olympic pairs LP. Using the trimmed mean, Sui and Han outscored Savchenko in PCS by 3.32 points. Using the untrimmed mean (all nine judges' scores), Sui and Han's margin changes to 3.41 points. The difference between discarding the highest and lowest scores and NOT discarding the highest and lowest scores is 0.09 points --less than a tenth of a point.

(Where Savchenko and Massot won was on TES, where they were almost 5 full points ahead of Sui and Han. in base value.)

Here are the numbers (untrimmed in parentheses).

Savchenko and Massot:

SS 9.21 (9.25), Tr 9.04 (9.03). Perf 9.32 (9.31). Ch 9.21 (9.22), Int 9.14 (9.17)

Factored total: 73.47 (73.57)

Sui and Han:

SS 9.57 (9.56), Tr 9.46 (9.44), Perf (9.54 (9.47), CH 9.71 (9.69), Int 9.71 (9.72)

Factored total 76.79 (76.98)

If you look at the range and distribution of the numbers on an actual protocol sheet it is instantly clear (at least, so it seems to me) that the numbers will ALWAYS come out like this, with no more than a fraction of a point overall difference between the trimmed and the untrimmed mean. Except, as I keep trying unsuccessfully to say , the truly outlandish outlier (0.75 instead of the intended 9.75 caused by a data entry error.

Furthermore, the Chinese judge was by no means always the highest for S&M or always the lowest for S&H. He was disciplined bu the ISU for accruing too many "anomalies" (scores outside the "corridor" as defined by the ISU judges' review procedure.) This is interpreted as showing apparent consistent bias or incompetence over many competitions and leads to penalties.

Anyway, bottom line, we are arguing over statistical conventions that amount to only tenths or hundredths of a point -- about the same order of magnitude as the rounding errors that arise in in computing the total PCS from the 5 individual components.

 

[Mathman]

The logic is similar with judging the students in the school/college.

I think this analogy is flawed. When a college student scores 87 on a standardized test, yes, that is what we are trying to find out -- how well has he mastered the material of the course.

But the purpose of figure skating judging is to select a student to crown Golden King for a Day (and two more, the Silver King and the Bronze King ). That is the only thing we ask of a judging system. (It would be cool, though, if the judging panel could say -- as in music competitions, for instance -- you all stunk; we won;t award the prize this year.)

To me, "giving information to skaters" is not the judges' job. A basketball referee is not expected to tell LeBron James that he lost the game because he missed all those shots.

 

[Baron Vladimir]

Well, you didn't count the GOE with it, and the scores from the FP? I remember i did the same analyses as you But I'm pretty sure the bias is proved by looking at the absolute scores of the judge. You can take away the scores of that (or any other) individual judge to see if it would change the results. Or add another of him/her for the same purpose. The trimmed mean is used not to let those scores affect the results, if may ?!
On the other topic, there is a difference between assessment of quality of the game and to say if the ball is out of the game or not. It is a totally different type of judges job with those two. I mean, in a basketball game you don't need the feedback to see why you won a game, you have your points on the table during the game based only on a fact if you score the point or not, not how well you do it.
And we are too much off the topic I guess, maybe people still want to discuss the topic of the tread...

 

[Skatesocs]

(It would be cool, though, if the judging panel could say -- as in music competitions, for instance -- you all stunk; we won;t award the preize this year.)

I'm conflicted between wishing this had been true for 2018 men worlds, or making sure the judging were better so that Kazuki Tomono had a deserved world medal.

 

[Blades of Passion]

Simply having the 5 component scores doesn't really provide feedback to a skater, especially with how the components actually get used (constantly misapplied and not differentiated enough). They have no idea why a judge scored them a certain way on each component. Only getting a detailed write-up or comment from a judge will give that clarity, same as under 6.0 scoring. Skaters under that system were constantly told to work on their stroking quality, speed, posture, cleanliness, expression, choreography, and musicality. They were given more feedback about artistic details back then, than they are now. The importance of body lines and performance and actual good choreography and interpretation is hardly a consideration at all anymore. It's devolved into a very basic game of "score as much as possible on your elements and put as many transitions as possible."

 

[CanadianSkaterGuy]

I agree that the components scores don’t give a whole lot of feedback about program aspects but 6.0 didn’t do that either. There were plenty of skaters who won simply because they did all the tricks the best even though they lacked refinement in their skating (Stojko and Bonaly come to mind)

 

[CanadianSkaterGuy]

Whom would you call the best in the field when it comes to skating skills, then? Kwan competed between 1994-2006 - we can say the best in 1994 was Yuka Sato, and best in 1995 was probably Lu Chen.

No I’m not saying Kwan didn’t have excellent skating skills - of course she does. I’m just saying her spiral sequence and her 2002 programs weren’t the best representation of her skating skills.

 

[CanadianSkaterGuy]

Slutskaya is absolutely right. She should've been first after the SP, which would've made her the overall winner. Kwan's 3F in the SP was underrotated, just watch it in slow-motion: https://youtu.be/K11MHq5ak78?t=137

I agree that it was UR and it went off the blade. It’s interesting how Scott Hamilton caught that thinking she “slipped off the pick” but the reality is her flip technique at the Olympics was horrendous, and she succumbed to it in the LP. Sasha’s flip Was lovely in the SP but she didn’t get the same spring in the LP and it was arguably two footed if you compare the jumps side by side you’ll see she doesn’t check her free leg out the same (her pointed toes are nice but so risky for leading to 2-foot landings which frequently were an issue for her). Of course Slutskaya’s 3F in the LP was a hot mess landing too.

Like Hughes’ 3-3 combos though Kwan’s 3F “looked clean” so I guess it was enough to put her 1st after the SP.

 

[Mathman]

Whom would you call the best in the field when it comes to skating skills, then? Kwan competed between 1994-2006 - we can say the best in 1994 was Yuka Sato, and best in 1995 was probably Lu Chen.

I think Arakawa could be mentioned, peaking in 2004.

 

[Mathman]

By the way, here is how you could catch conspiring judges and bloc judging back on the day. It is called the Spearman rank correlation test.

Let's say you want to measure the degree to which two judges are scoring in lock-step or in opposition to each other.

Skater.....Judge 1.....Judge 2 ....difference in rank (d) ... d squared

Sarah ........ 1 ..........3 ..................... 2 .......................... 4
Irina .......... 2 ......... 4 .................... 2 .......................... 4
Michelle ..... 3 ......... 2 ..................... 1 ........................... 1
Sasha ... .....4 ......... 1 .................... 3 ........................... 9

sum of d[sup]2[/sup] = 4+4+1+9 = 18.

Now do this:

r[sub]S[/sub] = 1 - 6*(sum of d[sup]2[/sup])/n(n-1)(n+1) = 1 - 6*18/4*3*5 = 1- 108/60 = -0.8.

This statistic ranges from +1 for identically matching rankings to -1 for exactly opposite rankings. In this example there is an 80% negative correlation between the rankings of the two judges.

 

[Baron Vladimir]

By the way, here is how you could catch conspiring judges and bloc judging back on the day. It is called the Spearman rank correlation test.

Can you prove with it that 'Western' and 'Eastern' block in judging really existed back in 6.0 days, would be my question for you I guess you can't prove if those are 'politically' involved decision more than matter of cultural preferencies, and we can leave that question aside. But I can bet that media was the one who invented block judging to sell figure skating to the common people, so i'm interesting in a question if it really existed statistically in a way media claiming it existed. Is figure skating was really a politics, or politic is used just to 'sell' figure skating?

 

[Mathman]

^ Unfortunately you can't really "prove" anything one way or the other, because the sample sizes are too small. (The distribution of this statistic is hard to describe mathematically, but is somewhat similar to the correlation coefficient r.) You can usually get just as good a feeling for things just by eyeballing the data

About bloc judging and judges making deals with each other, my impression at the time was that it was the ISU insiders themselves rather than the general news media that was pushing this narrative. Sonia Bianchetti was in the forefront of the movement to clear up the corruption from within. (She was kicked out of the ISU after running unsuccessfully for ISU president against the Cinquanta cabal.) One year (was it 1984?) the entire Soviet judging team was suspended for a year for being too blatant about "political judging."

Strangely, the ISU never asked itself whether the public perception that the sport was rigged and crooked might ever hurt it's popularity with the public.

As for cultural preferences, before the 1990s there was only one Soviet Union, so there was not a cluster of national federations that admired the supposed "Russian style" above others. The first cultural conflict of this sort came up back in the 19th century where the "British style" contended with (and lost out to) the Continental, or Vienna style. Strangely, the Vienna style was invented and popularized by the American champion, Jackson Haines. Haines' "fancy skating" became the rage in Europe (outside of England) but was rejected in the United States as being, well, a little too fancy.

 

[Baron Vladimir]

Well, that was my question, if ISU insiders through the media promoted figure skating the way they could, or the judges really behave that way? I mean, mistakes in judging will always exist, now how can you prove statistically back than they are made intenionally? Because in COP we saw ISU statistically 'proved' those situations (with 2 'unlucky' Chinese judges). I would rather believe in numbers, than what people say And i'm pretty sure those situations didn't hurt the popularity of figure skating among the general public. They actually made people to watch it (take Nancy-Tonya situation as an example). That all could only hurt figure skating in a context of being an Olympic sport, but oh well..

 

[Thrashergurl]

If Irina was 1st after the SP and Kwan 2nd, or 3rd....Kwan could’ve skated lights out and not made mistakes In the LP. It wouldn’t been less pressure going into the LP for her, and more on Irina. Irina’s LP was a hot mess as is, and Olympic pressure is no joke.

So much for the what if’s....it is what it is.

 

[Mista Ekko]

I would imagine if Irina was 1st after the SP she would've had a sharper FS,
Especially when she's admitting she felt biased against in this event,
It might've taken her mind off of that and made her feel like it's more attainable

But as you said, It's all speculation and the chaos theory prevails

 

[lesnar001]

I think it was more surprising that Michelle Kwan had problems in her long program than Irina Slutskaya.

IIRC, Irina had leads after the SP at 2000 Worlds and 2001 Worlds, and did not do as well in the LP.

If I had to guess, I think the pressure and expectations of the Olympics may have gotten to Michelle a little bit.
And probably would have happened even if she had been in 2nd place after the SP.
After all, being in 1st-2nd-3rd after the SP didn't really matter. If you win the LP, you win the Gold.

But who knows in actuality what would have happened in alternate scenarios.

I wonder if Sarah Hughes had been in the top 3 after SP, if it would have affected her adversely.
In other competitions, she seemed to tighten up in the LP if she was in contention for a title.
This mostly happened at US Nationals, although I think a couple of times at the GP series too.
She seemed to do better if she was coming from behind - like 2001 Worlds where she won Bronze after being 4th after the SP.

 

[Mathman]

Can you prove with it that 'Western' and 'Eastern' block in judging really existed back in 6.0 days, would be my question for you ...

I didn't give a very good answer to that question. Here is what I should have said.

The value of a statistic like the rank correlation coefficient is that it provides a way to quantify the degree to which two judges agree or disagree when they rank skaters. Without such a measure all we can do is mumble vaguely that the rankings of judges A and B seem to be sort of similar while judge C's rankings are somewhat different.

Obviously we can not expect a mere number ("80% correlation") to make any contribution to questions like, did the cold war increase interest in skating, are figure skating judges a pack of incompetent fools, should Slutskaya have won the 2002 Olympics, etc. A number is just a number.

 

[Baron Vladimir]

So in 6.0 time there was no any statistical method to prove 'biased' judging except people rumours...

 

[Skatesocs]

So in 6.0 time there was no any statistical method to prove 'biased' judging except people rumours...

? Mathman showed you a method.

 

[Baron Vladimir]

? Mathman showed you a method.

Kind of. I guess the problem with 6.0 is that you as a judge are giving only one number (or two numbers) per skater comparing to a set of numbers you are giving in COP. So it is harder to prove anything, statistically.