"Tatiana Tarasova outraged at refereeing of the World Championship of Figure Skating"

[Leonardo]

LOL, Chan's fans are the only people in the world who think his victory (in the LP especially) was fair, except maybe the (cheaters)judges and himself . And anyone who doesn't think this way (commentators, coaches, other skaters, audience, all the other fans) is not rational, doesn't understand the sport and so on. So funny Ridiculous really. But Takahashi is such a hero, even with the cheating, the lower pcs, still got first in the WTT and a world record.

 

[skatinginbc]

If we assume that these data represent measurement along a continuum...

But it won't make any difference in conclusions even if we use the formula: Median = L + W x (N/2 - C)/F
where:
L: the lower class boundary of median class
W: the width of median class
N/2: the sample size divided by 2
F: the frequency of median class
C: the previous cumulative frequency of the median class

Joubert PE median = 8.31 = Hanyu PE median
Joubert CH median = 8.31 = Hanyu CH median
Joubert IN median = 8.45 > Hany IN median = 8.33

Joubert median total > Hany median total (Yes, I should have said Joubert got a higher median than Hanyu in IN and tied with Hanyu in the other two categories).

Was there a French Judge on the panel?

No. The judges were: Evgeni ROKHIN (UZB), Karin EHRHARDT (Austria), Ekaterina SEROVA (BLR), Eddy WU (TPE), Hailan JIANG (CHN), Ebru ANILDI (TUR), Sung-Hee KOH (KOR), Elena FOMINA (RUS), Inger ANDERSSON (SWE).

That 9.50 that Joubert got for interpretation stands out like a sore thumb. It is 2.34 standard deviations above the mean.

Are you assuming a normal distribution in scores since the concept of standard deviation and mean is used in your identification of outliers? If the judges were calibrated to apply similar criteria in assessing the performance, and if no more than, say, 1-point discrepancy among judges was the threshold of their training qualification, then it was Hanyu's 9.50 (i.e., an absolute deviation of 1.25 from the median) that would be considered a rogue score. It was likely contaminated because the judge apparently did not apply the same criteria as other judges and it had a "gaming" effect in a small sample size--able to pull up the favorite by saving the adjacent high score from trimming (if the highest and the lowest are trimmed) or by inflating the non-robust standard deviation so that some high scores would fall below the cutoff (if the trimming is determined by standard deviation). Even if we assume a normal distribution of the scores and ignore the limitation of the sample size and the unknown distribution of measurement error, why do we use μ + 2σ (95.45% confidence interval) rather than μ + 3σ (99.73% confidence interval, which is employed by the International Tchaikovsky Competition as the criterion for identifying rogue scores)? Roughly 5% of the scores (or 1 in 20 observed scores) will fall outside the 2-standard deviation cutoff, but those "outliers" may still very well be part of the "normal" distribution. Why do we automatically reject a score greater than 2SD rather than take it as a possibly legitimate opinion from the judge and winsorize it to the nearest acceptable score?

Of course, if we assume a normal distribution and use Dixon's Q test, the 9.50 that Joubert received would look even worse: Q = gap/range = 1/1.5 = 0.67 > Q99% = 0.598

 

[skatinginbc]

LOL, Chan's fans are the only people in the world who think his victory (in the LP especially) was fair, except maybe the (cheaters)judges and himself .

I personally don't think there was cheating going on. There were so many judges from small federations that day (e.g., Uzbekistan, Taiwan, Turkey, Korea, to name just a few). I suspect some of the judges might be less experienced and thus scored the performances based on previous competition results to keep themselves within the prescribed corridor--"Reputation judging" in other words.

 

[Leonardo]

I personally don't think there was cheating going on. There were so many judges from small federations that day (e.g., Uzbekistan, Taiwan, Turkey, Korea, to name just a few). I suspect some of the judges might be less experienced and thus scored the performances based on previous competition results to keep themselves within the prescribed corridor--"Reputation judging" in other words.

I kinda agree with you. But the more experienced judges certainly do the same thing. The thing is, PChan has been falling all over the place for the whole year (since his victory at 2011 worlds), and still always get the highest PCs of all the guys (and all time biggest pcs), including performance (except today) for programs that sometimes are very obviously very far from brilliant: rushed, uninspired, w/ falls, problems with timing, etc. And of course, that are certain skaters who were better in CH, PE and IN in some competitions, and Chan won in every aspect. I DON'T like conspiracy theories, but it's hard not to think there's some serious cheating going on.

 

[Mathman]

Are you assuming a normal distribution in scores?

I did a few chi squared (not chi square ) goodness of fit tests, and it looks like that is an OK assumption. But the sample size (9) is so small that it is hard to conclude anything.

Q = gap/range

Yeah, it's that big gap that makes the 9.5 stand out.

..................8.50
..................8.50
..................8.50
8.00............8.50
8.00...8.25...8.50.....................................9.50

 

[CAS]

Disagree with Doris about Tat's politiking. She was doing this IMO in the entire article which I read in translation. She is the consummate politician.

She is the best. Although she also reacts emotionally to a performance so I think she genuinely preferred D&W's FD and knowing her it's not a surprise to me she prefers the music of Die Fledermaus over Funny Face. She also squeals like a teenager when Joubert skates regardless of how he skates She´s a legend but she´s also human and thinks with her heart over her head at times.

But back to politics. When this season began I had a dream. Just a dream, absolutely nothing real that might have been whispered about or hinted by any skating officials (judges, federation heads, those sorts of officials). I dream a lot

In this dream fans all over message boards believed D&W and V&M were in a league of their own, that no one could touch them. I even had coaches in my dream saying the same thing. BUT in another scene there were whispers. Whispers that D&W were in a league of their own, that V&M were on par with P&B and P&B could at least beat V&M and fight for Gold by Worlds. Such a disconnect between what was being said out loud in my dream and what was being whispered in my dream. I thought why V&M? If so many believe and say out loud that these two are equals and miles in front of anyone else then why are the people whispering in my dream targeting one of them as suddenly inferior, not only to their main rival but to those below them.

And inside that dream I had another dream where I tried to understand this movement. I first rationalized that the Olympics are two years away and some people might want to shake the foundation of the leaders and this was the year to start. Why one team over the other? Perhaps one team has already had the greatest Olympic success AND the girl has a history of injury AND history shows us that when an ice dance team starts to slide they tend to run and save face siting "injury" as a reason to skip the rest of the season. So if I whisper these doubts to others I'm friendly with they go into a competition with prejudices suddenly seeing things that might not be there. So stage one of my preparation for 2014 is to take down one team this year and it seems to be working, they are getting rattled and becoming obsessed with the numbers and their standing and they start tinkering with things which cuts into their preparation and they flounder around not knowing what to do to improve. And if I'm lucky they take the rest of the season off OR get so flustered completely mess up the rest of their season. But I picked the wrong team OR these two teams don't react as ice dance teams of the past did and turn out to be tough competitors. They continue on to compete at 4CCs and Worlds where fewer of my whispering friends are working and those damn rookies become charmed with a refreshed team who overcame their insecurities and with what they see because they unfortunately haven´t been around much while I was whispering.

I look forward to my dreams for next season because I think the whisperers are not happy right now but I'm curious who they will be whispering about next year. .

 

[Mathman]

If the judges were calibrated to apply similar criteria in assessing the performance, and if no more than, say, 1-point discrepancy among judges was the threshold of their training qualification, then it was Hanyu's 9.50 (i.e., an absolute deviation of 1.25 from the median) that would be considered a rogue score.

Here is how the ISU actually does it. (ISU Communication # 1631, paragraph E). For GOEs:

For each element the "average" score is computed. This is the mean of all the 9 judges (no trimming), together with the score of the referee, counted twice, together with the scores of any members of the ISU Officials Assessment Committee who might be present. For each judge, for each element, compute the absolute difference of that judge's score from the average. Now add these discrepancies up for the whole program.

If the total is off by more than the number of elements, that counts as an anomaly.

So for instance in the short program there are seven elements. If all the other judges give -3 GOE on an element and you give +3, you are still OK if you are really close on all the other elements.

For program components a judge is even more comfortable because the plus and minus discrepancies cancel instead of accumulate. So if the mean for each of the five components is five across the board and you give 0.00, 0.00, 0.00, 10.00, and 10.00, you are well within the corridor. (For PCS you are allowed to be off by a total of 7.5 points. In this example the judge is off by only 5.)

 

[emdee]

CAS
You got it. the whisperers have to end with I&K at the top so whisper all their way to the top. Some of the whisperers are shouting instead of whispering!!

 

[Mathman]

Whispers that D&W were in a league of their own, that V&M were on par with P&B and P&B could at least beat V&M and fight for Gold by Worlds.

That is very interesting, because the whispers I heard were just the opposite. That Virtue and Moir had finally eclipsed Davis and White once and for all (we had a thread with that very title) and that Weaver and Poje had a good chance to get at least silver if not gold in 2014.

I remember the days when there was real funny stuff with the scoring...Thank goodness those days are gone.

Thank goodness.

 

[blue dog]

CAS, Mathman, and emdee--I am calling Leonardo di Caprio! You three need an inception!

 

[dorispulaski]

Or is that an intervention

 

[blue dog]

Doris, they're past the point of intervention. It's time for an inception!

 

[dorispulaski]

I do understand that it's never too late to start a new beginning...

And what would you like them to begin ?

I'd vote for cynicism classes

 

[blue dog]

I do understand that it's never too late to start a new beginning...

And what would you like them to begin ?

I'd vote for cynicism classes

Or, I'd settle for an inception-themed ice dance program.

 

[Dragonlady]

Or, I'd settle for an inception-themed ice dance program.

I'm trying to imagine what the costumes would look like.

 

[CAS]

I'm trying to imagine what the costumes would look like.

http://www.google.ca/imgres?q=spinn...135&tbnw=144&ndsp=29&ved=1t:429,r:2,s:277,i:9

 

[dorispulaski]

They could dress as a couple of chicks emerging from Easter Eggs.

They could be wearing hats like the one Fabian is wearing:

http://washingtonexaminer.com/files/styles/article-main-image/public/file5InVCn

 

[bestskate8]

I kinda agree with you. But the more experienced judges certainly do the same thing. The thing is, PChan has been falling all over the place for the whole year (since his victory at 2011 worlds), and still always get the highest PCs of all the guys (and all time biggest pcs), including performance (except today) for programs that sometimes are very obviously very far from brilliant: rushed, uninspired, w/ falls, problems with timing, etc. And of course, that are certain skaters who were better in CH, PE and IN in some competitions, and Chan won in every aspect. I DON'T like conspiracy theories, but it's hard not to think there's some serious cheating going on.

TAT just confirmed that in her statement. :thumbsup:

 

[skatinginbc]

I did a few chi squared (not chi square ) goodness of fit tests, and it looks like that is an OK assumption.

Aha. It reminds of our little debate in another thread about whether the "measurement" approach can assume a normal distribution of the scores.

Here is how the ISU actually does it....

By saying "if no more than, say, 1-point discrepancy among judges was the threshold of their training", I meant "the usual distribution of measurement error also influences the judgment on whether a suspected score can be seen as contaminated and thus justifiably trimmed." Otherwise, the scores, despite of falling in the tail of the bell curve from the perspective of standard deviation, are still an integral part of the distribution and convey meaningful opinions from the judges.

 

[Mathman]

(T)he scores, despite of falling in the tail of the bell curve from the perspective of standard deviation, are still an integral part of the distribution and convey meaningful opinions from the judges.

I agree with this conclusion. But I think the point of trimming the mean is not for any statistical purpose, but rather to take the edge off blatant bias and nationalism on the part of the judges.

Every judge has to give an inflated score to skaters from that judge's country. The reason why is that all the other judges are doing it, so that is the only way to make it fair.