Armstrong Doping Evidence an Obvious Fabrication

In article <[email hidden]>,

(Larry Weisenthal) said:

It is 99% certain that the present charges against Armstrong are bogus.

Most of us just take Armstrong at his word.

[email hidden] (Larry Weisenthal) a écrit :

Quoted message said:

It is 99% certain that the present charges against Armstrong are bogus.

YOUR calculation is bogus.

There are around 150 urine controls done by the lab on the TdF every
year.

They don't take samples to ALL the competitors. They only take samples
to the first competitors, plus some other randomly chosen.

Armstrong won the TdF in 99, had the yellow jersey most of the time and
won 4 stages. It is completely logical that he was the most controlled
competitor, and had the most samples in the lab.

-- Olivier

Silver0l said:

[email hidden] (Larry Weisenthal) a écrit :

Quoted message said:

It is 99% certain that the present charges against Armstrong are bogus.

YOUR calculation is bogus.

There are around 150 urine controls done by the lab on the TdF every
year.

They don't take samples to ALL the competitors. They only take samples
to the first competitors, plus some other randomly chosen.

Armstrong won the TdF in 99, had the yellow jersey most of the time and
won 4 stages. It is completely logical that he was the most controlled
competitor, and had the most samples in the lab.

-- Olivier

Does the lab have any remaining sample from Armstrong?

I'm sure that I remember it was about 30 total tests on the 1999
specimens, of which 6 were from Armstrong and 6 others from a single,
unnamed rider. As I write this I can't put my finger on the reference,
but read below:

The point is that the lab has only run a total of 150 rEPO tests in its
existence. It only collected about 150 urine specimens from this years'
riders and tested only a random sampling of those. So I'm guessing
that the tests are technically a lot of work and/or expensive, so I'd
be surprised if the research study in question on the 1999 specimens
included more than 30 (which, again, would be 20% of ALL the tests the
lab has ever done).

Let's say that they collected 150 urine specimens in 1999 (the same as
in 2005). I'll grant that they could have been skewed toward the race
leaders. Armstrong said that he provided 17 specimens (see quote
below). So about 10% of the total specimens collected for testing were
those of Lance. 6 specimens from Lance were chosen for testing. 11
specimens were not chosen. Let's say that the second rider was also
someone really prominent and he submitted 17 specimens also. Anyway,
let's say that there were 150 total specimens and Armstrong submitted
17 and 6 were randomly selected and the other guy also submitted 17 and
6 of his were selected, meaning that for everyone else there were 117
submitted and 18 selected for testing. Doing a 3 x 2 table gives a two
sided p value of 0.018. Meaning that there is only a 1.8% chance ("98%
certainty," to use Olivier's terminology) that such a skewing of
specimens for only two riders could have occurred by chance. And I
think that this is a very conservative example. The more total urine
specimens actually submitted, or the fewer of the second rider's
specimens actually submitted, the less likely it is that these
specimens represent a random selection.

Quoted message said:

Quoted message said:

Quoted message said:

http://www.bicycling.com/article/0,3253,s1-13111,00.html?category_id=441

<excerpts from above>

Armstrong says he was tested 17 times during the 1999 Tour; the lab has
run, according to Ceaurriz, 150 total tests, including samples from the
1998 Tour, (with, he says, a number of positive signs of EPO use) but
he did not know if all of Armstrong's samples were tested. If not,
there remain more which could be tested, but the protocol problems
remain.
The basis for the claim: According to l'Equipe, 12 samples returned
positive results for EPO; six from Armstrong, six from unidentified
riders. France Soir, a French daily newspaper, reported last week that
the other six samples all belonged to one rider and that his name was
being withheld by "all those capable of revealing (it)." France Soir
further alleged that the rider in question must have been quite
successful at the Tour and had become "too popular to end his career
dishonestly."

Ressiot told us he never pursued the other six samples, and admitted he
concentrated on Armstrong. "Why? Quite simply because he was such an
immense champion but one that had so much suspicion around him and so
much mystery," he said, adding that he had so far held off writing much
about Armstrong simply because he felt only direct proof would be
acceptable. "I don't have all the numbers of all the Tour riders in all
stages from the 1999 Tour," he said, pointing out that he worked for
months to get just one set of numbers. "That is very complicated
because there are roughly 200 riders in for nearly 20 stages and the
numbers for each rider's form are different each day." Ressiot said at
this point, it would be hard for him to get the numbers to match to the
other six samples but that "it is something the UCI could do if they
wanted."

Quoted message said:

Quoted message said:

Meaning that there is only a 1.8% chance ("98%

certainty," to use Olivier's terminology) that such a skewing of
specimens for only two riders could have occurred by chance. <<

P.S. And the above calculation is merely based on the chance that such
a distribution (6/11, 6/11, and 18/117) would have fallen out
completely randomly. Factor in the fact that it was not only skewed,
but _positively_ skewed, in the same direction, toward two _specific_
individuals, who happened to be the two most prominent people in the
race, and the probability that this occurred by random chance becomes
even more remote.

No, it is obvious that the specimens in question were SELECTED for
testing by someone in the UCI lab who knew the identity of the
specimens before the testing took place.

- Larry Weisenthal

I don't see your point, Larry. Are you saying the lab is suspect
because it hasn't run enough samples in its time? Wasn't the research
study meant to determine whether new tests could be used on old
samples? I don't see where the lab says how many tests they performed
in the study. I only see your guess of 30. Aren't your calculations
all based on that guess?

(Larry Weisenthal) said:

I'm sure that I remember it was about 30 total tests on the 1999
specimens, of which 6 were from Armstrong and 6 others from a single,
unnamed rider. As I write this I can't put my finger on the reference,
but read below:

The point is that the lab has only run a total of 150 rEPO tests in its
existence. It only collected about 150 urine specimens from this years'
riders and tested only a random sampling of those. So I'm guessing
that the tests are technically a lot of work and/or expensive, so I'd
be surprised if the research study in question on the 1999 specimens
included more than 30 (which, again, would be 20% of ALL the tests the
lab has ever done).

Let's say that they collected 150 urine specimens in 1999 (the same as
in 2005). I'll grant that they could have been skewed toward the race
leaders. Armstrong said that he provided 17 specimens (see quote
below). So about 10% of the total specimens collected for testing were
those of Lance. 6 specimens from Lance were chosen for testing. 11
specimens were not chosen. Let's say that the second rider was also
someone really prominent and he submitted 17 specimens also. Anyway,
let's say that there were 150 total specimens and Armstrong submitted
17 and 6 were randomly selected and the other guy also submitted 17 and
6 of his were selected, meaning that for everyone else there were 117
submitted and 18 selected for testing. Doing a 3 x 2 table gives a two
sided p value of 0.018. Meaning that there is only a 1.8% chance ("98%
certainty," to use Olivier's terminology) that such a skewing of
specimens for only two riders could have occurred by chance. And I
think that this is a very conservative example. The more total urine
specimens actually submitted, or the fewer of the second rider's
specimens actually submitted, the less likely it is that these
specimens represent a random selection.

Quoted message said:

Quoted message said:

>

http://www.bicycling.com/article/0,3253,s1-13111,00.html?category_id=441

<excerpts from above>

Armstrong says he was tested 17 times during the 1999 Tour; the lab has
run, according to Ceaurriz, 150 total tests, including samples from the
1998 Tour, (with, he says, a number of positive signs of EPO use) but
he did not know if all of Armstrong's samples were tested. If not,
there remain more which could be tested, but the protocol problems
remain.
The basis for the claim: According to l'Equipe, 12 samples returned
positive results for EPO; six from Armstrong, six from unidentified
riders. France Soir, a French daily newspaper, reported last week that
the other six samples all belonged to one rider and that his name was
being withheld by "all those capable of revealing (it)." France Soir
further alleged that the rider in question must have been quite
successful at the Tour and had become "too popular to end his career
dishonestly."

Ressiot told us he never pursued the other six samples, and admitted he
concentrated on Armstrong. "Why? Quite simply because he was such an
immense champion but one that had so much suspicion around him and so
much mystery," he said, adding that he had so far held off writing much
about Armstrong simply because he felt only direct proof would be
acceptable. "I don't have all the numbers of all the Tour riders in all
stages from the 1999 Tour," he said, pointing out that he worked for
months to get just one set of numbers. "That is very complicated
because there are roughly 200 riders in for nearly 20 stages and the
numbers for each rider's form are different each day." Ressiot said at
this point, it would be hard for him to get the numbers to match to the
other six samples but that "it is something the UCI could do if they
wanted."

Quoted message said:

Quoted message said:

SELECTED for

testing by someone in the UCI lab<<

.... meant to say "SELECTED for testing by someone in the French lab..."

(Larry Weisenthal) said:

Quoted message said:

Quoted message said:

Meaning that there is only a 1.8% chance ("98%

certainty," to use Olivier's terminology) that such a skewing of
specimens for only two riders could have occurred by chance. <<

P.S. And the above calculation is merely based on the chance that such
a distribution (6/11, 6/11, and 18/117) would have fallen out
completely randomly. Factor in the fact that it was not only skewed,
but _positively_ skewed, in the same direction, toward two _specific_
individuals, who happened to be the two most prominent people in the
race, and the probability that this occurred by random chance becomes
even more remote.

No, it is obvious that the specimens in question were SELECTED for
testing by someone in the UCI lab who knew the identity of the
specimens before the testing took place.

Did they say the total number of tests they performed, or is your
calculation based on your guess of 30, which you don't restate above.
That's very bad form, Larry.

Quoted message said:

Quoted message said:

don't see your point, Larry. Are you saying the lab is suspect

because it hasn't run enough samples in its time? Wasn't the research
study meant to determine whether new tests could be used on old
samples? I don't see where the lab says how many tests they performed
in the study. I only see your guess of 30. Aren't your calculations
all based on that guess? <<

I didn't make that number (30) up. I remember that from something I
read over the past 24 hours, only I just don't remember where. I'm
sure that Olivier can confirm or refute this rather easily; so I'm not
trying to pull the wool over everyone's eyes. What makes this
(small/30) number very credible is that the lab has performed only 150
rEPO tests in its entire history, including research studies and doping
control studies, since the inception of the test -- circa Y2K. So I'd
be astonished if the total number of specimens tested in the "Armstrong
Sting" was more than 30 (and, again, I remember reading that this was
the number).

So you've got one rider (Armstrong) with 6 of those specimens and
another rider (as yet unidentified) with another 6 and that leaves only
18 "randomly" selected for everyone else in the race. Again, Olivier
says that 150 urine specimens were donated (which squares with what I
read somewhere else for this year's total). And Armstrong claims that
he donated 17 specimens (just over 10% of the total of 150 specimens
taken from among 200 riders who started the race), which also seems
about right. Now, presuming that the mysterious "2nd doper" was also
selected to donate 17 specimens, and 6 were tested and found to test
positive, the simple (contingency table) statistics shows that there is
only a 1.8% probability that this would have occurred by chance alone.

But this calculation is probably very conservative (meaning that there
is much less than 1.8% probability that the specimens were randomly
selected), for the following reasons:

1. It assumes that every single Armstrong specimen selected for testing
tested positive. Olivier says that rEPO can only be detected for 3 days
after it is administered. I explained earlier why it was very unlikely
that TdF dopers inject themselves over the entire 3 week course of the
race (RBCs last for 120 days and RBCs stimulated by rEPO don't appear
for more than a week after it is administered). So if one or more
ADDITIONAL Armstrong specimens were included in the study but tested
negative, then the probability that they were chosen randomly (from
among the 150 specimens collected) is even more remote.

2. It also assumes that every single "Mystery Doper" specimen tested
positive (but see #1).

3. It assumes that "Mystery Doper" was tested just as often in 1999 as
was Lance. Who from that race would you guess would have been selected
for testing as many times as Lance? I can't think of anyone. So if
"Mystery Doper" submitted fewer than the 17 specimens submitted by
Lance, the chance that his 6 positive specimens were chosen randomly
becomes even more remote and the total probability again drops much
below 1.8%.

4. I was conservative in using a 2-sided "P value." It would have been
justifiable to try to (a priori) reject the null hypothesis that the
Armstrong specimen wasn't intentionally selected for testing, meaning
that it would be justifiable to use a 1-sided "P value." Using a
1-sided P value would reduce the probability that it could have
occurred by chance to just 0.9% (using Olivier's cavalier terminology;
this means that it is "99% certain" that the specimens of Armstrong and
the "Mystery Doper" were pre-selected for testing by someone in the lab
who knew their identity in advance).

5. As explained earlier, the fact that the results were skewed in the
SAME (positive) direction for each of a particular set of two riders
(the ones who were disproportionately selected for testing in the first
place) makes the probability even more remote that this was a chance
occurence.

- Larry Weisenthal

(Larry Weisenthal) said:

Quoted message said:

Quoted message said:

don't see your point, Larry. Are you saying the lab is suspect

because it hasn't run enough samples in its time? Wasn't the research
study meant to determine whether new tests could be used on old
samples? I don't see where the lab says how many tests they performed
in the study. I only see your guess of 30. Aren't your calculations
all based on that guess? <<

I didn't make that number (30) up.

Until we know the actual number, your selection of 30 is suspect. they
may have tested all of them. Or, upon finding one positive test for a
specific rider, they may have gone back and deliberately tested more
samples for that rider. I don't know what protocol they used. Their
purpose was not to detect cheaters but to test their own methods, so
once they have a positive sample, they would want to verify the
method, not the rule violation.

Quoted message said:

I remember that from something I
read over the past 24 hours, only I just don't remember where. I'm
sure that Olivier can confirm or refute this rather easily; so I'm not
trying to pull the wool over everyone's eyes. What makes this
(small/30) number very credible is that the lab has performed only 150
rEPO tests in its entire history, including research studies and doping
control studies, since the inception of the test -- circa Y2K. So I'd
be astonished if the total number of specimens tested in the "Armstrong
Sting" was more than 30 (and, again, I remember reading that this was
the number).

So you've got one rider (Armstrong) with 6 of those specimens and
another rider (as yet unidentified) with another 6 and that leaves only
18 "randomly" selected for everyone else in the race. Again, Olivier
says that 150 urine specimens were donated (which squares with what I
read somewhere else for this year's total). And Armstrong claims that
he donated 17 specimens (just over 10% of the total of 150 specimens
taken from among 200 riders who started the race), which also seems
about right. Now, presuming that the mysterious "2nd doper" was also
selected to donate 17 specimens, and 6 were tested and found to test
positive, the simple (contingency table) statistics shows that there is
only a 1.8% probability that this would have occurred by chance alone.

But this calculation is probably very conservative (meaning that there
is much less than 1.8% probability that the specimens were randomly
selected), for the following reasons:

1. It assumes that every single Armstrong specimen selected for testing
tested positive. Olivier says that rEPO can only be detected for 3 days
after it is administered. I explained earlier why it was very unlikely
that TdF dopers inject themselves over the entire 3 week course of the
race (RBCs last for 120 days and RBCs stimulated by rEPO don't appear
for more than a week after it is administered). So if one or more
ADDITIONAL Armstrong specimens were included in the study but tested
negative, then the probability that they were chosen randomly (from
among the 150 specimens collected) is even more remote.

2. It also assumes that every single "Mystery Doper" specimen tested
positive (but see #1).

3. It assumes that "Mystery Doper" was tested just as often in 1999 as
was Lance. Who from that race would you guess would have been selected
for testing as many times as Lance? I can't think of anyone. So if
"Mystery Doper" submitted fewer than the 17 specimens submitted by
Lance, the chance that his 6 positive specimens were chosen randomly
becomes even more remote and the total probability again drops much
below 1.8%.

4. I was conservative in using a 2-sided "P value." It would have been
justifiable to try to (a priori) reject the null hypothesis that the
Armstrong specimen wasn't intentionally selected for testing, meaning
that it would be justifiable to use a 1-sided "P value." Using a
1-sided P value would reduce the probability that it could have
occurred by chance to just 0.9% (using Olivier's cavalier terminology;
this means that it is "99% certain" that the specimens of Armstrong and
the "Mystery Doper" were pre-selected for testing by someone in the lab
who knew their identity in advance).

5. As explained earlier, the fact that the results were skewed in the
SAME (positive) direction for each of a particular set of two riders
(the ones who were disproportionately selected for testing in the first
place) makes the probability even more remote that this was a chance
occurence.

- Larry Weisenthal

<[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

P.S. And the above calculation is merely based on the chance that such
a distribution (6/11, 6/11, and 18/117) would have fallen out
completely randomly. Factor in the fact that it was not only skewed,
but _positively_ skewed, in the same direction, toward two _specific_
individuals, who happened to be the two most prominent people in the
race, and the probability that this occurred by random chance becomes
even more remote.

No.

Randomness, by its very nature, is immune to outside influence.

jtaylor said:

<[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

P.S. And the above calculation is merely based on the chance that such
a distribution (6/11, 6/11, and 18/117) would have fallen out
completely randomly. Factor in the fact that it was not only skewed,
but _positively_ skewed, in the same direction, toward two _specific_
individuals, who happened to be the two most prominent people in the
race, and the probability that this occurred by random chance becomes
even more remote.

No.

Randomness, by its very nature, is immune to outside influence.

The probability *is* lower for the case where the pair of six samples
is for two specific names than for for the case where the two names
can be any riders in the set.

"Martin Smith" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

jtaylor said:

<[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

P.S. And the above calculation is merely based on the chance that such
a distribution (6/11, 6/11, and 18/117) would have fallen out
completely randomly. Factor in the fact that it was not only skewed,
but _positively_ skewed, in the same direction, toward two _specific_
individuals, who happened to be the two most prominent people in the
race, and the probability that this occurred by random chance becomes
even more remote.

No.

Randomness, by its very nature, is immune to outside influence.

The probability *is* lower for the case where the pair of six samples
is for two specific names than for for the case where the two names
can be any riders in the set.

That it not what Weisenthal said. You should re-read the post above, noting
specifically the following part:

"...probability that this occurred by random chance becomes even more
remote."

The probability of a random event does not change if the resulting choices
happen to be interesting to a human observer.

In article <[email hidden]!nnrp1.uunet.ca>,

jtaylor said:

"Martin Smith" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

jtaylor said:

<[email hidden]> wrote in message
news:[email hidden]...

> P.S. And the above calculation is merely based on the chance that such
> a distribution (6/11, 6/11, and 18/117) would have fallen out
> completely randomly. Factor in the fact that it was not only skewed,
> but _positively_ skewed, in the same direction, toward two _specific_
> individuals, who happened to be the two most prominent people in the
> race, and the probability that this occurred by random chance becomes
> even more remote.
>

No.

Randomness, by its very nature, is immune to outside influence.

The probability *is* lower for the case where the pair of six samples
is for two specific names than for for the case where the two names
can be any riders in the set.

That it not what Weisenthal said. You should re-read the post above, noting
specifically the following part:

"...probability that this occurred by random chance becomes even more
remote."

The probability of a random event does not change if the resulting choices
happen to be interesting to a human observer.

Yes it does. In one case we ask for twelve samples, six for one rider
and six for another. If there are 30 samples from five riders, where
there are six samples in the set for each of the five, then there are 5
sets to choose from and we want 2 sets. 5 take 2 is 10, so there are ten
possible ways to choose 2 sets from 5 sets.

But if we specify which sets we must have, then there is only one
combination that works.

Quoted message said:

Quoted message said:

I don't see your point, Larry. Are you saying the lab is suspect

because it hasn't run enough samples in its time? Wasn't the research
study meant to determine whether new tests could be used on old
samples? I don't see where the lab says how many tests they performed
in the study. I only see your guess of 30. Aren't your calculations
all based on that guess? <<

Firstly, the "30" wasn't a guess. I read/heard that somewhere (I'm
thinking maybe BBC at 2AM local time or some such thing). I'm sure
that Olivier can rescue us on this one by coming up with the precise
number.

But what we do know is that the lab has only run 150 rEPO tests in its
entire history; which includes tests for research purposes, tests for
quality control purposes, and tests for doping control purposes. So
this makes the figure of "30" for the present "Armstrong Sting" cohort
highly plausible (it's really just a check/validation on my memory of
something I'm sure I read/heard). The point is that the lab didn't
test 150 specimens and 6 of them happened to be Armstrong's and 6
happened to be "Mystery Doper." Instead, the lab tested a very small
number of specimens (I maintain that this will, indeed, be shown to be
"30"😉 and a grossly disproportionate number came from Armstrong and
from "Mystery Doper."

Meaning that someone in the Lab intentionally selected the Armstrong
and Mystery Doper specimens for testing. In the context of the
certainty that someone in the laboratory leaked the results to the
reporter, this points strongly in the direction of foul play,
particularly as someone in the UCI also intentionally leaked the
information and because of the virtual certainty that the UCI people
have long been in communication with the lab people.

- Larry W

"Martin Smith" <[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Quoted message said:

The probability of a random event does not change if the resulting

choices

Quoted message said:

Quoted message said:

happen to be interesting to a human observer.

Yes it does.

Perhaps you believe is intelligent falling as well?

Quoted message said:

In one case we ask for twelve samples, six for one rider
and six for another. If there are 30 samples from five riders, where
there are six samples in the set for each of the five, then there are 5
sets to choose from and we want 2 sets. 5 take 2 is 10, so there are ten
possible ways to choose 2 sets from 5 sets.

Ignoring the number of samples per rider, the situation resolves to a one in
10 chance of _any_ combination of two riders being chosen. Whether one of
them is Armstrong or not, the probability remains the same.

Quoted message said:

But if we specify which sets we must have, then there is only one
combination that works.

The above is tautological and does not support your position that human
interest can influence chance.

<[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Quoted message said:

Quoted message said:

The probability *is* lower for the case where the pair of six samples

is for two specific names than for for the case where the two names
can be any riders in the set. <<

Just in case jtaylor has plonked Martin's responses, I'm re-posting it.
Martin is correct in his defense of my explanation.

You agree with him when he states the following?:

">> The probability of a random event does not change if the resulting
choices

Quoted message said:

Quoted message said:

happen to be interesting to a human observer.

Quoted message said:

Yes it does."

If two things are selected by chance from a set of five, then the
probability that the selection will match a pair of names we have previously
drawn (by whatever means - ouija board, or Martin's latest sock-puppets, for
example) is the same no matter which pair of names we have already in hand.

Quoted message said:

I've explained my reasoning. Firstly, I didn't make up the numer "30."
I read/heard it. I just can't remember where.

I believe that you are honest about it. But it does not mean that
30 is correct, since I have no idea how reliable your forgotten source is.

In article <[email hidden]!nnrp1.uunet.ca>,

jtaylor said:

<[email hidden]> wrote in message
news:[email hidden]...

Quoted message said:

Quoted message said:

>The probability *is* lower for the case where the pair of six samples

is for two specific names than for for the case where the two names
can be any riders in the set. <<

Just in case jtaylor has plonked Martin's responses, I'm re-posting it.
Martin is correct in his defense of my explanation.

You agree with him when he states the following?:

">> The probability of a random event does not change if the resulting
choices

Quoted message said:

Quoted message said:

happen to be interesting to a human observer.

Quoted message said:

Yes it does."

If two things are selected by chance from a set of five, then the
probability that the selection will match a pair of names we have previously
drawn (by whatever means - ouija board, or Martin's latest sock-puppets, for
example) is the same no matter which pair of names we have already in hand.

When the two names are specified, there is only one combination that
matches, ie there are only two sets of six samples that can be drawn
from the five available sets. When the names are not fixed in advance,
there are ten combinations that work. You can have any pair of sets of
six samples.