Shelby's side of the story from wapo

The boring truth is that doping is right throughout professional sport now. But most aren't caught because doping is always ahead of antidoping.

Case in point. You are just one of the public that wants to believe.

Maybe you find these kind of vague emotional appeal non-quantitative arguments persuasive, but without reliable quantities and confidence intervals and percentages, it just means whatever you want to believe -- interpretation is in the mind of the beholder.

Tell me, just how big is "throughout" and "most"? And why all this vague talk about "professional sport" -- I don't care about doping in sports like bodybuilding or weightlifting or baseball, or events like sprints and field events; what about distance running? The answer? You don't really know. In other words, the foundation for your beliefs is ignorance, with a little bit if anti-doping incompetence. The larger the ignorance, the more room for your beliefs.

I suppose a bigger reason is that most dopers are not even tested, as they are failing to place on the podium, and not in the OOC registered pools.

The goal of anti-doping is not to catch all the dopers -- that is too unrealistic. They prioritize testing the important ones as the ones that are winning races. But even then, anti-doping can still be effective by acting as a deterrent, and in parallel, educating the athletes.

But for sure it doesn't help anti-doping authorities catch more dopers by accusing and railroading innocent athletes to 4-year sanctions.

I wasn't making an argument - I was stating a fact. Anyone who has followed the sport for years, observed it, and become aware of what antidoping experts have concluded knows this. But that obviously excludes you. Renee Ann Shirley, the Jamaican whistleblower, said some years ago that doping "is in all sports at the top levels and in all countries, with the collusion of sports governance bodies". Then David Howman says doping is always "more sophisticated than antidoping" - which is why it is impossible to stop. Add to that the confidential athlete surveys that show doping is much more prevalent at championship level than testing reveals. All of that just opens the door to the reality of doping in sports. But it is a door you will not go through. Your life would lose all meaning were you to do so.

"But for sure it doesn't help anti-doping authorities catch more dopers by accusing and railroading innocent athletes to 4-year sanctions".(quote)

Yeah - like Shelby. You are a joke.

Nevertheless, your fact is so vague that it is effectively meaningless and borders superfluous. It's not that I don't want to go through that door, but that the door leads nowhere interesting, except to an imaginary land where we can imagine the worst based on what we do not know because it is not well quantified. I do not find it interesting to merely say doping exists in "sport", or that anti-doping isn't sophisticated enough (so give us more money).

Your facts are either non-quantitative, or the quantity is such a wide range, it is not possible to conclude whether the problem is small, medium, or large, so you fill in the blanks with what you want to believe.

Renee Anne Shirley is effectively saying doping exists, i.e. is greater than 0% -- not a big revelation. What can this Jamaican tell us specifically about distance running today? David Howman told us once that prevalence is more than 10% (in cycling after the Armstrong conviction). Again, not a big revelation. Confidential survey results have not yet been proven to be reliable or accurate. Published results from the very same event, using 95% confidence intervals, tell us prevalence ranges between 9.7% to 47.9%, with +/- 30% estimated survey non-compliance. In any case, mixing up men and women and strength and speed and throwing and jumping and distance running tells us very little about distance running.

Comparing or estimating doping prevelance with testing results (1-2%) never made sense, as no one suggests doping prevalence is 1-2%. This also catches more dopers than you might first think, as athletes are tested multiple times, sometimes 10-20x in one year.

If you include 1 sigma (are you using 2?) and 30% noncompliance, the upper limit of that 2011 survey is actually 70ish % dopers (including PEDs and forbidden methods but excluding offenses more than 12 months prior) at worlds.

With 2 sigma, 75% or so. Just to be complete - confidence intervals and noncompliance effects go both ways.

The injustice of "strict liability" goes way beyond the single case of Houlihan, as anyone who follows anti-doping closely will be aware. Hajo Seppelt found a swimmer, a judoka, a handballer, and an entire women's field hockey team put on the defense of having to prove how substances entered their body without their knowledge, all at great personal cost, reputational cost, and a significant interruption in their career, and in one case, losing a championship medal that they could not get back.

As we have learned from many public cases, the innocent athlete, in order to prepare a proper defense, must hire a lawyer, a private investigator, and pay a lab to conduct testing. There is often a significant delay, so it may not even be possible to collect sufficient evidence, that may be have long been consumed or discarded. The athlete must also have enough luck in still having relevant samples for testing, and finding a sympathetic testing authority and/or anti-doping panel.

I think this is a lot to ask of an innocent athlete, especially from poor athletes from poor countries where English or French may not be their native language, and the necessary resources (money, lawyers, labs) are simply not available.

While you sing the praises of a one sided process created by one-side, which you consider right because that side won't change it, I find myself again in agreement with German Athletes ("Athleten Deutschland") who said in an official media release "In the future, it must be made easier for athletes to prove their innocence. For example, they could receive increased assistance in proving their innocence and enhanced legal aid. Those who are innocent should receive redress for the harm inflicted and the suffering they have endured. In addition, the investigative and analytical capacities of anti-doping organizations must continue to be strengthened."

And I also find myself in agreement with the German NADA anti-doping officials who say "The principle of 'strict liability' does not release anti-doping organizations from providing full evidence of the doping violation." and "Anti-doping organizations are obliged to fully prove a violation." and "As the central regulator, WADA is now called upon to make direct requirements for dealing with the findings of the (sabotage) Study and the effects on the doping control system worldwide."

Looks like you are mixing up a few things.

I didn't use any sigmas, but went with the published 95% confidence intervals.

These confidence intervals were not functions of non-compliance, but rather a measure of something else, like the expected spread of distributions of birthday questions versus the calculated probability.

There wasn't just one survey, but two.

In one survey, the published 95% CI range was 39.4%-47.9%. There was no non-compliance measurement as such, but there was a recommendation to remove the fastest 30%, based on measured response times, speculated to be too hasty to be reliable, in order to improve reliability.

When following the study recommendation, the published 95% CI range dropped significantly to 26.5%-36.3%. It clearly only went one way when they corrected for hasty responses.

In the second survey, the published 95% CI was 9.69%–32.7%, and the non-compliance was measured, with 95% CI, to be 26.28%–37.52%. This prevelance estimate has already factored in the non-compliance, so there can be no further adjustment.

I already went both ways, providing the best and worst from both surveys, rounding to 9.7% - 47.9%.

You also misunderstand the main point. The point was that these 95% CI ranges are far too broad, especially when considering the large proportion of non-compliance, to consider that any survey method can give a reliably accurate prevelance estimate for that championship event. In effect, the interpretation is that only within this broad range, can we be 95% confident that the "true" prevalence will be contained within that range.

If you want to now suggest I should consider 70% and 75% as the upper end of 95% confidence, you are only making it worse. If I said that based on two survey methods, we can be 95% confident that 2011 WCA prevelance was between 9% and 70-75%, that is only marginally better than saying it is somewhere between 0 and 100% based on no data at all.

To be reliable, survey methods need to be improved to minimize survey non-compliance, or at least the effect of survey non-compliance.

This is actually a good point. Not exactly precise.

FYI 63.4% result when correcting for 30% non-compliance (which you suggested earlier) of the dopers in that 39-47% survey; presumably that results in a 95% CI of 56 - 70%.

So 10% - 70% for these two studies? The lower limit is basically what Army's favorite WADA guy said (at least 10% he said iirc), but now we have an upper limit too.

Thank you.

Some corrections.

I did not suggest to correct for 30% non-compliance in either survey, nor did the authors. Such an exercise would require knowing the type and magnitude of each form of non-compliance -- say with 95% confidence.

In the first survey, there was only one recommendation, based on response times, and not on any form of non-compliance. The authors recommended to disregard the fastest 30% of responses based on an empirical observation that the fastest responses were inexplicably and significantly inconsistent with the slowest 70% of responses, suggesting many automatic "yes" responses from athletes who didn't want to take the survey, but also didn't want to say no when asked to support a doping study.

We cannot assume nor conclude with any confidence that all forms of non-compliance were removed by this recommended exercise. The remaining 70% may still have significant non-compliance unknown to the authors.

The authors did not suggest any further exercises to correct for any non-empirical theoretically calculated distortions, such as "under-reporting doping" (leading to 63.4%), and we cannot place any confidence on any result derived from non-empirical calculations.

In the second survey, there was no recommendation for any corrections -- just a best fit model that estimated both prevelance and non-compliance.

But again, using two survey methods to estimate prevelance within a range of 10% to 70% (95% confidence boundaries), is hardly an improvement from our starting point with no knowledge or data.

Survey methods simply are not yet proven to be reliably accurate, and need to be refined. This was the recommendation from both studies. Survey methods are promising, but in need of refinement.