Contador's defense is perfectly sound and logical: if it can be established that the clen positive couldn't be the result of a blood transfusion (or any other doping practice), then yes, he should walk away unscathed. The Sherlock Holmes defense is perfectly fine - the alternative would be punishing riders even if we knew for a fact they were completely innocent (no fault or negligence of their own).
Since no evidence is ever going to be 100%, I think, as I posted on another thread, that it amounts to a balance. The likelihood of transfusion vs. the likelihood of meat. Based on what has come out so far, it seems highly unlikely that he ate contaminated meat, but can he show that it is even more unlikely that he transfused blood with CB in it? He can’t and shouldn’t be expected to rule out blood transfusion absolutely, but if he can show that it is more unlikely than contaminated meat, I agree that is the best he can do.
The issue should be whether Contador has actually proved that a blood transfusion must be ruled out as the origin of the clen. The RFEC says he has, and we won't know until we see the doctor's report but we have reason to suspect that's bogus. That's what you guys should be criticizing, not the Sherlock Holmes defense which makes perfect sense.
Agreed. I think LMG hit the nail on the head. If Bert can show he tested negative for CB in June, during the period when he most likely would have transfused, that is strong evidence against transfusion, maybe strong enough to counteract the lack of evidence for contaminated meat. IMO, a clean passport during this period is not strong enough evidence that he didn’t withdraw blood. The two recent cases not withstanding, a passport test can be beaten.
That's not what they meant. EU standards say that 1% of cattle should be tested in connection with clen. That's the standard. If I recall correctly (please correct me if I am wrong here), the amount tested in the region was less than 100, which if my math is correct is far less than 1% of the cattle in the region (assuming there are 850,000 cattle in the region, the 100 represent .0012%). So it doesn't matter how many cattle were tested in OTHER regions of Spain, if they can establish that very little testing was done on a relatively large population. So while on the whole it is relatively improbable to eat clen tainted meat in the EU, without knowing the sampling error for the particular region, you cannot make that same argument. There are still farmers using clen in their livestock, that much is certain (recall there was a positive case out of all of the EU testing and then recent arrests related to clen and livestock and athletes too).
There is no mention in the RFEC report of a 1% testing standard, as far as I can see. If you see it there, or found a link for it somewhere else, I would be interested. However, I can assure you it’s not necessary. In order to obtain a significant value--with an error of 1%, or any other value desired--the ONLY THING that matters is the ABSOLUTE NUMBER of cattle tested--NOT the % of the total tested. This is why, in electoral polls, one can query only one thousand voters out of millions or hundreds of millions, and still have a fairly low error. It's why the EU can test only 0.25% of its cattle--as Bert's team pointed out after the RFEC decision, without ever mentioning a 1% standard. And it’s why Bert’s team said, in the RFEC report: “For best results from a statistical point of view (i.e., with a confidence level of 95% and a prediction error of 1%), 8,586 cattle should have been analyzed in 2007,which contrasts with the 97 cattle actually analyzed that year.” Again, it doesn’t matter whether that 8586 is 10%,1% or 0.1% of the total cattle present, the error is THE SAME. (See
http://www.rogerwimmer.com/mmr/mmrsampling_error.htm
for a calculator that determines the error at 95% confidence limits for different size samples. If you plug in 8586, you get 1.06%).
In the following statement, they say, “with the sample of animals tested, the probability of identifying animals in the Basque Autonomous community contaminated with clenbuterol is extremely low, 0.001221 in 2007.” I frankly don't know what this means, the passage doesn't make sense to me. It might be saying that the 97 animals tested is 0.1221% of the total animals in that region. IOW, there are about 80,000 cattle in the entire region. If you or someone else can make better sense of this, let me know, all I can say is my problem here is not because it's in Spanish.
But it really doesn't matter, as far as I can see. What matters is the 97. This corresponds to a higher error, about 10% (see the link I provided above. Also note that there is a power relationship involved in error estimates. That is, to reduce the error by two, you must increase the number sampled by four. 97 is about 100x lower than 8586, which means the error is about ten times higher).
To understand these statistical measurement, it might be helpful to think of a coin toss. If you toss a coin, the probability of getting a head should be 50%. If you toss the coin ten times, the number of heads you get will be within some error of 50%, or five heads. If you toss the coin one hundred times, the error will be smaller, that is, the number of heads will usually be closer to 50%. If you toss the coin one thousand times, the number of heads will be still closer to 50%. The more tosses, the smaller the error.
Now suppose you toss the coin ten times, and repeat the ten times a number of times. In 95% of these repeats, the number of heads will be within the error. E.g., the error is about 30% or 5 heads plus or minus 1.5 heads. If you toss the coin one hundred times and repeat many times, in 95% of the repeats, the number of heads will be within some smaller error, 10% (50 heads plus or minus 5 heads). For one thousand tosses, it will be 500 plus or minus 15 heads.
It's basically the same with cattle testing. So when Bert's team says that 8586 cattle must be tested for a confidence level of 95% and a prediction error of 1%, they mean that if you tested a random sample of 8586 cattle over and over, 95% of the time the number of contaminated cattle would be within 1% of the value actually obtained. So if one hundred cattle were contaminated in the test of 8586, you could conclude with 95% certainty that the % of contaminated cattle in the entire population corresponds to 99-101 per 8586.
And their further point is that when you test just 97 cattle, the result you get has a 95% certainty for only a much larger error--in this case, about 10%. So if ten cattle out of the 97 tested positive, the 95% confidence limits wqould be 9-11. Note that that is still a fairly small % of the 97 cattle. And since in fact zero cattle tested positive, the 95% confidence limits involve a much smaller number of cattle. I'm not sure how many, but suppose there was one contaminated steer in the 97. You can conclude with 95% confidence that the proportion of contaminated cattle in the entire population is no more than 1.1/97, or about 1.1%. So obviously, even only 97 catttle indicate that there is a very low probability of contamination.
And remember, this calculation concedes that Bert's argument about the 97 being too low has validity. In fact, by the usual rules of sampling, it does not. By the usual rules, you assume that the % of contaminated cattle is determined by the tests of the entire area, Spain, unless you have an a priori reason for believing that the Basque region is different. But even allowing them this argument, they still can't come up with a number that suggests a very strong possibility of contamination.