Chinese Skaters' Age In Question

[Mathman]

I'll wait and see what Mathman will say. We may ask him to give up his handle should he give us a wrong answer.

*Ahem* Strange to say, my last PhD student wrote his dissertation on strategies for dealing with missing data in the setting of investigations in the social sciences. Statistically, the problem with just filling in all the blanks with the same dummy number (the mean of the actual responses, for instance), is that this artificially lowers all the variances, making the study appear to be more robust than it really is. Results will satisfy standard criteria for statistical significance, when in reality they are baloney.

(This consideration is swamped, however, by the fact than people simply lie on questionnaires. )

 

[Violet Bliss]

*Ahem* Strange ..........missing data ......problem ........ blanks .........dummy number .........artificially .............. in reality they are baloney.

people simply lie

In other words, there are lies, damn lies, and statistics, which are baloney.

 

[skatinginbc]

*Ahem* Strange to say, my last PhD student wrote his dissertation on strategies for dealing with missing data in the setting of investigations in the social sciences. Statistically, the problem with just filling in all the blanks with the same dummy number (the mean of the actual responses, for instance), is that this artificially lowers all the variances, making the study appear to be more robust than it really is. Results will satisfy standard criteria for statistical significance, when in reality they are baloney.

(This consideration is swamped, however, by the fact than people simply lie on questionnaires. )

Haha, another beating-around-the-bush answer from Dr. Mathman, as his wont. In this particular case, substituting the questionable data with 0 will increase, not lower, the variances, and make the study appear to be weaker, not more robust, than it actually is. And worst of all, it will artificially lower the mean, which makes the statistical result rather off the mark. Anyway, back to the subject, I was thinking that some Chinese skaters might have falsified their dates of birth but forgotten to change their start dates for skating and, as a consequence, caused the average start age or even the mean of variances to differ somehow. My preliminary study however shows no clear evidence to support my hypothesis.

 

[prettykeys]

Age 0 should be a neutral assumptive substitute for statistical purposes, no?

Haha, another beating-around-the-bush answer from Dr. Mathman, as his wont. In this particular case, substituting the questionable data with 0 will increase, not lower, the variances, and make the study appear to be weaker, not more robust, than it actually is. And worst of all, it will artificially lower the mean, which makes the statistical result rather off the mark.

skatinginbc, you are misunderstanding Mathman.

First of all, substituting "0" for "-6" as a starting age will DECREASE the artificially higher variance (as opposed to using the raw reported value of -6.) 0 is in no way a neutral substitute, however, because it still artificially brings down the mean or the median. Something like substituting a mean would be theoretically better than 0, but...

Mathman was explaining what happens when one substitutes the mean of the responses for missing or blatantly erroneous data (such as starting skating 6 years before one is born). The mean may not be altered as much (which is good), but it would also artificially lower variance.

*Ahem* Strange to say, my last PhD student wrote his dissertation on strategies for dealing with missing data in the setting of investigations in the social sciences. Statistically, the problem with just filling in all the blanks with the same dummy number (the mean of the actual responses, for instance), is that this artificially lowers all the variances, making the study appear to be more robust than it really is. Results will satisfy standard criteria for statistical significance, when in reality they are baloney.

(This consideration is swamped, however, by the fact than people simply lie on questionnaires. )

 

[Violet Bliss]

I was joking. Sigh.

How can age -6 or 0 be taken seriously for starting skating? GIGO applies to statistics too, which I think was Mathman's point as well. :think:

 

[Mathman]

'Wait a minute! How can that girl still be eligible for juniors? I know for a fact that she has been skating for twenty years!"

"True, but she's only 14."

 

[skatinginbc]

skatinginbc, you are misunderstanding Mathman.

Actually, it is my fault not making my point clearer. I should have said, "In this particular case, substituting the questionable data with 0 AS OPPOSED TO EXCLUDING THE DATA FROM THE CALCULATION will increase, not lower, the variances, and make the study appear to be weaker, not more robust, than it actually is."

Mathman's information was correct of course. It is just that I expected him to specify the best way to treat erronous data in this particular case. There were two options presented: (1) Excluding the data, (2) Filling in with 0. By mentioning filling in with the mean (which was not one of the options mentioned in our prior discussions), was he suggesting that it was the best way? I guess he did not. But if he did, how is it better than simply treating the data as non-existence given that it is assumed to be a random data entry error?

 

[Mathman]

I think I will just say that this is a question of mathematical substance in the application of statistical methods to the social sciences, and one that is the subject of ongoing theoretical and applied research. If anyone wants to learn something about it, the place to start is with the "Expectation-Maximization Algorithm" (bearing in mind that there have been many new developments since the 1970s).

As for the skater who began skating at age minus six, I beleive her best competitive season was when she was minus two, turning minus one. After that she had that tremendous growth spurt and it was all down hill from there.

 

[SeaniBu]

It is just that I expected him to specify the best way to treat erroneous data in this particular case. There were two options presented: (1) Excluding the data, (2) Filling in with 0. By mentioning filling in with the mean (which was not one of the options mentioned in our prior discussions), was he suggesting that it was the best way? I guess he did not. But if he did, how is it better than simply treating the data as non-existence given that it is assumed to be a random data entry error?

I think that says it all right there, doesn't it? Missing data is not data, is it speculation or data because we had to speculate?

It is data regardless.

 

[skatinginbc]

I think I will just say that this is a question of mathematical substance in the application of statistical methods to the social sciences....

I was in my wanton, naughty mood. After I read my own postings again, I felt they were not that funny at all. I am sorry if I have offended you or been disrespectful in any way.

Thanks for your answer.

 

[Mathman]

^ Oh goodness, not at all. I love wanton and naughty. Where would this board be if we weren't wanton and naughty sometimes?

I greatly enjoy your contributions to the forum.

 

[Mathman]

I think that says it all right there, doesn't it? Missing data is not data, is it speculation or data because we had to speculate?

It is data regardless.

The biggest problem for researchers in the social sciences comes when they are trying to investigate correlations between different variables. For instance, suppose one of the questions on the questionnaire is, "What is your annual income?" and another question is, "Do you have freckles?" We might want to know whether people with freckles make more money than people without. Indeed, maybe the purpose of the study is to develop a list of factors that will allow us to predict someone's income (on the average) if we know a lot of other things about them.

But many people will not answer the question about income (or will not answer truthfully, which introduces another kind of problem). So what should we do? If we throw out the entire questionnaire when one datum is missing, then we are discarding a lot of data that are relevant to our study. If this happens often, then we will not be able to conclude anything at all and we will have wasted our time and money doing this project in the first place.

Maybe the most interesting finding will turn out to be that people with freckles do not share information about their income as readily as do people without.

One thing we could try is this. Take the people who have freckles (and share a bunch of other characteristics as well) and who do report their income, and then use some kind of average of that to stand in for the incomes of people with freckles who choose not to report their income. Well, it is easy to see many problems with that approach.

The most successful method seems to be some kind of iterative "bootstrap" idea, where we make original assumptions based on all other factors in the study, run the new numbers through the computer, then revise our assumptions on the basis of the shape of the new data and do it all over again.

Needless to say, when the subjects of the study are real people, any kind of mathematical analysis is fraught with peril and very likely dead wrong.

 

[Bluebonnet]

Has anyone checked skaters' height records (including USA)? They are a big mess. The records could be messed up intentionally and/or unintentionally. Just saying...

 

[Johnet]

Has anyone checked skaters' height records (including USA)? They are a big mess. The records could be messed up intentionally and/or unintentionally. Just saying...

Does this even comparable with age issues? Height has nothing to do with rules and it changes over time, while birthday should stay the same.

 

[Bluebonnet]

Does this even comparable with age issues? Height has nothing to do with rules and it changes over time, while birthday should stay the same.

No, there's no rules about height. What I had in mind was not those teenagers who are still growing and changing heights. I was talking about 20+ adults. I've seen many of them have different records in different sites and resources. Those are messy too. Of course, heights are not as sensitive as age. Just want to throw it out here. That's it.

 

[antmanb]

No, there's no rules about height. What I had in mind was not those teenagers who are still growing and changing heights. I was talking about 20+ adults. I've seen many of them have different records in different sites and resources. Those are messy too. Of course, heights are not as sensitive as age. Just want to throw it out here. That's it.

I think the thing with height is that it can be subjective - first thing in the morning straight out of bed everyone will be a little bit taller than at the end of the day when you've been on your feet all day. Barefoot, wearing shoes, wearing trainers, wearing skates....many different ways to get height wrong. Birth date, not so much, birth dates that allow you to compete in competitions that you othewise would be excluded from....just too much coincidence.

Also I don't buy the errors, if they really were errors then why aren't there more glaring errors there? Why aren't numbers transposed, why is it only an error of a couple of years older or younger. The biggest data entry errors are transposed numbers/letters.

Ant

 

[Johnet]

No, there's no rules about height. What I had in mind was not those teenagers who are still growing and changing heights. I was talking about 20+ adults. I've seen many of them have different records in different sites and resources. Those are messy too. Of course, heights are not as sensitive as age. Just want to throw it out here. That's it.

The mess is easy enough to explain, aside from what antmanb said, it's very probable that some website record athletes' height a few years ago and never bother to update it. Could this ever happen to birthdates? -- birthdates should NOT be updated or changed. It's not even about sensitivity, they are totally DIFFERENT and same with antmanb, I don't buy the errors.

 

[feraina]

I have followed the whole drama closely, from the first post on FSU that used translate.google.com to find discrepant birthdate records of athletes on the Chinese federation website (which was indeed before the AP article, which pointed to the same document).

I have been following Chinese skating forums and watching Chinese figure skating competitions for years, and my personal perspective is that age-falsification among Chinese figure skating athletes follows a certain pattern: young skaters from provincial teams (usually pre-teen) burst onto the national scene showing some precocious talent, the media rave over them and exclaim how young and promising they are. They get noticed by the federation at the national level, and suddenly their ages increase just before they get sent to their first international competition. (I have only seen this with female skaters; it doesn't necessarily mean that none of the male skaters' ages are deliberately falsified, just that they may have happened earlier in their careers before they got anyone's attention) At the same time they get noticed by the national team, they also have to move far away from wherever they are from, hundreds/thousands miles away from their parents, and sent to live in national team controlled dorms. This is especially tough for the young girls, who don't even get to go home for Chinese new year, which is the biggest holiday (it's like Thanksgiving, Christmas, and Easter combined) and nationally decreed to be a holiday for everyone (so either the federation is violating national law to make them train through that holiday or the laws make a special exception for young athletes!). In any case, it is at a time when these young children (and often only children) are removed from parents' sphere of influence and protection that the age change takes place. And from then on, the coaches and the team officials, with vested interest in their succeeding in international competitions, become also the parental figures in their lives. Does anyone remember that Tong said in an interview after he revealed his romantic relationship with Pang (from an article I translated) that he didn't know whether to pursue her or not until a top-dog team official gave him the stamp of approval, saying that later in life Pang will be the only one who will be able to fully share/understand the happiest and saddest moments of his skating life, that he found the courage/determination to tell her his feelings, and moreover that he saw that team official as the father figure in his life? Tong is a grown man in his 30's. If he is under such strong personal influence of the team officials, just imagine what the little 10-, 11-, 12-year girls' mental world is like.

One little story: when little 13-yr-old Zijun Li was first entered into Asian Winter Games over Chinese New Year, she said she wished she could see her parents instead, but hoped all the hard work was worth it and will eventually pay off. But then the host nation discovered she was underage (age rules are the same as for GP's), she was left in Beijing in the empty training center, with her coach and training mates gone, and still not allowed to go home for Chinese New Year. To me, it just shows that the federation and team officials don't care about athletes' welfare at all. She was already really bummed that all that hard training was for nought, at least they could've let her go home to see her family.

I really believe that age falsification is happening with the knowledge, permission, and even encouragement of high-level skating federation officials -- they get promotions and financial bonuses if their athletes perform well at major competitions, and if the cheating is discovered, the athletes get to take all the blame. It's the Chinese version of the "win I gain, lose you pay" U.S. financial system -- the big investment bank bosses get the huge bonuses if their reckless gambles pay off, and the American taxpayers pay for the losses if the gambles don't pay off. So the big bosses in both instances have all the power to make all the decisions, saddle someone else with all the problems when things go awry, and therefore every incentive to cheat and be reckless.

So I think some penalty is indeed necessary in the case of the current age falsification scandal. But I think it should firstly be done in a fair way that would be the same punishment meted out to another country found out with the same conduct, and secondly should be done in a way that should not exacerbate the "win I gain, lose athletes pay" situation in Chinese figure skating. And taking away athletes' previously won medals, banning all athletes from competitions for a year or two, while leaving the administration officials alone, is precisely the wrong thing to do! They could always just blame the athletes for all the age discrepancies and sit pretty in their bureaucratic perches and wait for the next young crop of athletes to turn up for them to mold and exploit as they see fit, or maybe they have already made career and personal advancements based on the success of underage athletes, and are now out of the skating world altogether.

Just as it would seem totally unfair to try to correct the U.S. financial system by punishing taxpayers/voters severely for the poor investment strategies of Fannie Mae/Fred Mac/Goldman Sachs -- after all, taxpayers are also voters who sort of control the white house and capitol hill, who legislate and execute regulations that ultimately oversee the financial entities, so you could say taxpayers/voters have to experience the punishment in order clean up the financial system. And in a way, this is slowly happening, with the grass root supported passage of the Obama financial regulation bill (the biggest achievement of his administration so far, IMHO), which, among other things, stipulates that big bosses cannot reap a large chunk of their bonuses until years later, when it is proven that their investment strategy really pays off in the long term instead of reaping short-term gain and causing long-term problems to the financial entity. So what will perhaps improve the Chinese skating federation is grass root supported reform of the administration that makes it more transparent, and the officials responsible for problems (like the discovery of age falsification) as well as the gains (competitive success). And obviously that will take time. Look how many years it took to get this financial regulation bill passed, and how many more it will take to get it actually implemented, and we live in a democracy where the people actually have a voice. In China, for the athletes, their families, and their fans to finally garner enough support to change the political culture due to direct punishment of athletes, it will definitely not take place very quickly, certainly too late to save the skating career of some athletes.

This is why I think that the belief that meting out severe punishments to the athletes (by banning them from competition for years, especially if their age isn't even in question) is both unfair to them, and also ineffective -- it is misguided to think that within that authoritarian culture that athletes' suffering will quickly, if ever, lead to a change in the political culture.

 

[ImaginaryPogue]

Very articulate.

My fundamental question though is how to do you punish this system without punishing the athletes? I mean, it seems you suggest that we let reform take place from the inside, but the fact is hardworking athletes around the world suffer as a consequence of said cheating, regardless of the directing mind.

 

[bigsisjiejie]

I basically agree with feraina's description of how the falsification happens. However I disagree on the following points:

--The rather forced analogy between the Chinese skaters' age situation and the US financial situation. Sorry, the former is actually a relatively simple set of conditions and the latter is not. Resolution of the former is relatively simple though highly unpalatable to both the protagonists (Chinese) and the international "guardians" (ISU). Resolution of the latter is not.

--The idea that change will take a long time. Baloney. Things in China happen really really quickly as soon as somebody near enough the top of the pecking order plants his foot firmly in someone's tushy and so it goes down the line. If the Chinese wanted to, they could get this Federation cleaned up and get the skaters' ages back to reality within a few months. Definitely by next season. Yes, it means some lumps have to be taken and probably medals returned. And there is no question of loss-of-face for the Federation, at the very least.

--Grassroots change? Don't bet on it. In the first place, WHO in China IS the grassroots--that are also independent and outside the current system? WHERE are these people who will show up and champion the cause of truth and fairness? Some anonymous people on a Chinese forum or BBS? The average Chinese fan simply doesn't care that much, and even if they did, wouldn't know how to go about effecting change. For that matter, I'm at a loss to figure out how one would effect a bottoms-up change in the current context. This is still a top-down society in almost every aspect, skating included.

--How can you be fair to every other national Federation that sticks to the age limit rules and give the Chinese a free pass? By playing the "poor young Chinese athletes as victims" card? Or giving the time to straighten themselves out and fly right? When the athletes get picked young and agree to keep going in this system, they become part of the collective and part of the bargain is to be the vehicle for national glory through international competition wins, in return for the Federation being responsible for complete financial support. I believe these athletes are not completely blameless in carrying on this charade, though I acknowledge they are caught between a rock and a hard place. But you know what? So are a lot of the rest of us in life.

My bottom line: There is simply no way to properly punish the chief instigators--who are not the athletes--without at least some athletes becoming collateral damage. I still think the only measure that has a prayer of being taken seriously by the Chinese and get them to force permanent changes, is a full suspension of the entire Federation, for at least one year, from senior and junior international competitions. Athletes, coaches, judges, officials. I believe the ISU has this possibility already in their bylaws. As for the Chinese sports apparatus, they would do well to start cleaning house in the Skating Federation. I see this last bit as a distinct possibility this year, with or without ISU sanctions, as this has become a national embarrassment in sporting circles.