Nobody talking about Ross Tucker v Paula radcliffe on twitter.... spat central!

Deez, is this Lauren Fleshman? The same LF that, when she discovered she had a "breathing problem," went to a world renowned doctor, err, Alberto Salazar for help with said "breathing problem?" Ever since that info came out, I've found LF to be the hippest of hypocrites, bar none. Truly flabbergasting that she can try to "call out" people with a straight face, yet had no trouble running to Alberto for medical "help." Trying to help Paula by creating a straw man out of identity politics is sad, even for LF's low standards.

dznuts wrote:

IMO Tucker made Radcliffe look pretty dang bad. Then Fleshman comes in and goes completely off topic calling out Tucker for "mansplaining". Tucker has been coming at male dopers hard for years. Nothing his said to Paula had anything to do with her being female.

Also love when he asks about Actovegin and Thyroid abuse in sport (to which he gets no answer). The lack of discussion among elites about these things makes me feel like it's all way more common than we would think.

She really is a pleasant piece of work. Gary please explain.

Bobble Head wrote:

[quote]Walsh fan wrote:

Hey people, what ever happened to all the blood bags taken from Dr. Fuentes office during the doping raid? Isn't it funny how that happens (along with the raid and arrest of Jama Aden) and years later no action is taken, nothing mentioned. The perps can bribe the judges to do nothing and the public then forgets.

Perhaps the Spanish police are utterly useless? Perhaps they're all like Manuel from Fawlty Towers?

I loved his question about Actovegin.

Good discussion. I guess the thing about the ABP and discussion of the1/1000 probability and 'outliers' is that in some ways it's missing the point - the ABP is personal, for each athlete and built up over time using data points obtained from the athletes blood samples. The more samples, the more data, the more reliable the ABP will be. Natural variations will occur so that one athlete having an off-score of 95 and one having an off-score of 75 means nothing without context.

Paula was perhaps THE most tested athlete of the mid 2000s and so if she released her data we would see her blood tests throughout this period and how she reacted to altitude, when she was sick etc. Of course though anything pre-2009 is pre-ABP and therefore deemed to be 'invalid'. (The irony is that the ABP was formed on the basis of blood scores obtained in the mid-2000s, so they trusted the data enough to do this......)

It's also interesting with what you write about the transparency of the panel used in the WADA ABP operating guidelines. Check out page 49 of the document (

) where it specifies the process of what happens after a suspicious score is obtained. The question with Paula is that we don't know where in the process her dodgy score got to in 2012. We know that she refers to 'experts' (plural) 'clearing' her, so this means we know that the first expert considered her sample to be 'likely doping' - note: despite what she'd written on the test form about being at altitude and sleeping in an altitude tent etc.

But then we don't know whether this was a stage c (where it first gets seen by a panel of 3 experts, including the initial expert) or stage e (where they put together a 'documentation package') or stage f/g, where the athlete is informed and has to come up with reasons why the score could be so high. (This is pretty much where Froome currently is).

THAT would be interesting to know. I rather think it got to stage f/g and suddenly there was a desperate scramble to find reasons to exonerate her. (That's based on an interview she did when she mentioned about the result coming back to British Athletics and them saying how it wouldn't be a problem, they would explain that she was at altitude and also injured at altitude and not crushing red blood cells)

Finally, and going back to the experts, we don't know who they were, and whether the 'vote' was 2-1 or 1-2 or whatever. It could just have been 1 expert that got her off the hook. Which then leads to questions about possible corruption. The IAAF was rife with corruption at this point. Big money was changing hands to cover up positive tests. Are we sure that no 'experts' weren't also bribed to change their minds about tests? Or that the IAAF exerted pressure to make sure positive tests got buried. That was pretty much happening anyway when we read the emails from Nick Davies and the Saugy references.

Well, here's the thing, the document says it's an ATPF if the result is outside of the 99th or 99.9th percentile for the expectations of THAT individual. 1/100 or 1/1000 may be different for me and for you, but it's still a probability of 1/100 or 1/1000 in either case! If you do a hundred tests, you're likely to get a 1/100 reading. Over the number of drug tests (Runner's World says about a dozen per year for top athletes) and number of statistical tests per drug test (markers, longitudinal tests, all the things you're looking at) and number of athletes being tested, there are going to be tons of false positives at the 1/100 or even 1/1000 level. Maybe the document is just worded imprecisely, perhaps they do make a correction for the total number of tests an athlete has taken over their life, I dunno. That would require an increase in the threshold of statistical significance as the athlete ages and the number of tests increases (making it easier to get away with doping). For some tests they say the information they learn makes it harder for them to get away with doping, so maybe that cancels out sometimes and the issue disappears in the end. I dunno. [It's late right now and I came back and finished this paragraph last, I have a feeling I'm going to wake up and see things differently in the morning]

You're right that more data will make it easier to catch people who are cheating. That's an increase in the true positive rate. But I'm trying to understand the false positive rate, which is the rate at which people are accused of cheating who aren't actually cheating.

Yeah your last paragraph is one of the things I was getting at. And it's not just corruption, but human fallibility. It's not clear to me what the human component brings to the table.

Good point about noting the 1/100 or 1/1000 refers to the individual, not the general population. Re: Paula, of course we don't know where the 2012 test shows up on the spectrum, as we have no idea about her ABP and the upper/lower limits of this. Given this, we also don't know where the 2003 and 2005 tests would fit in on this spectrum either - yes, of course they are 'invalid' as they were pre-ABP, but clearly they deemed suspicious enough at the time to prompt a lot of target testing.

Think you are massively overestimating the number of tests going on for elite athletes. On another thread we saw that for example, Sandre Moen was being blood tested on average 2 times a year, and it was stated that Paula had 14 blood tests through the IAAF between 2001 and 2008 (not sure if this includes tests carried out by other agencies, i.e. UKAD, Marathon majors etc.?). And for the Kenyans and Ethiopians? Don't forget there is NO blood testing in either of these countries and never has been, so I would imagine their ABP also consists of 1-2 blood scores per year.

In other words, when you have maybe 10-20 tests in a lifetime, then going over the 1/100 or 1/1000 levels is fairly suspicious. And remember the list of athletes leaked by Fancy Bears who were 'Likely Doping' or 'Suspicious Passport'. ALL of these people were over these limits, that's why they were on this list. And these are limits set very high to avoid false positives!!! You don't need to be master statistician to work out that something stinks.

Transparency is non-existent. And Paula doesn't want people digging into the details of her tests. I'm interested in digging it up because she is an icon in the sport, and her marathon time is iconic - and like Lance in cycling, you need people like this to be exposed for change to come. I'm actually very open to be persuaded otherwise and to consider the evidence, but first you need to be given the data.......

DHT123 wrote:

Think you are massively overestimating the number of tests going on for elite athletes. On another thread we saw that for example, Sandre Moen was being blood tested on average 2 times a year, and it was stated that Paula had 14 blood tests through the IAAF between 2001 and 2008 (not sure if this includes tests carried out by other agencies, i.e. UKAD, Marathon majors etc.?). And for the Kenyans and Ethiopians? Don't forget there is NO blood testing in either of these countries and never has been, so I would imagine their ABP also consists of 1-2 blood scores per year.

...

Correct and what happens is athletes wait until after they have an infrequent test and then they go hard on the sauce. The other thing is that in places where there is a lot of top athletes, there is no element of tester surprise. Everyone has a mobile these days.

[quote]dsfasdsadfasf wrote:

Well, here's the thing, the document says it's an ATPF if the result is outside of the 99th or 99.9th percentile for the expectations of THAT individual. 1/100 or 1/1000 may be different for me and for you, but it's still a probability of 1/100 or 1/1000 in either case! If you do a hundred tests, you're likely to get a 1/100 reading. Over the number of drug tests (Runner's World says about a dozen per year for top athletes) and number of statistical tests per drug test (markers, longitudinal tests, all the things you're looking at) and number of athletes being tested, there are going to be tons of false positives at the 1/100 or even 1/1000 level.

[quote]

That seems to be assuming that you take a blood test and you just get random values as the result. That is not the case in a real-world blood testing scenario.

DHT123 wrote:

Re: Paula, of course we don't know where the 2012 test shows up on the spectrum, as we have no idea about her ABP and the upper/lower limits of this.

...

Transparency is non-existent. And Paula doesn't want people digging into the details of her tests.

What's missing here so far, which demonstrates why Paula wants as little known about her data as possible:

1) In 2003, she didn't "just" cross the current thresholds, but more importantly her Hct increased by 30% within a few days, while the RET-% decreased.

2) Similar situation in 2005, but with a smaller jump.

3) Her off-score in 2012 was so high because of a very high Hct with normal RET-%.

ad 1+2: Those numbers came out over time. When Tucker did his calculations, he underestimated the 2003 change to be 22%, see

Note that even that less extreme change looked very suspicious. If you then consider the drop in RET-% (not known to Tucker back then), it is clear that that change cannot have been caused by dehydration/hemoconcentration/plasma shift.

Both can, however, be explained with blood transfusions.

This also demonstrates that neither Paula nor the IAAF can blame altitude for those values, because she didn't visit altitude between those data, but weeks before the 1st measurement.

Yet, the IAAF didn't mention any of the above, although it is public knowledge.

ad 3: Just as ludicrous. High Hct caused by altitude comes along with lower RET-%, and hers in 2012 were over 50% higher than in 2003.

Both, however, can be explained with using EPO at altitude.

Again, the IAAF didn't discuss this.

RE ABP: how generous these thresholds are, was demonstrated by Russian EPO cheats not even crossing those once. The original argument was to avoid false positives, yet crossing the thresholds only results in an investigation, not necessarily in a ban, so why not use better thresholds?

And yes, they originated from testing alleged clean athletes. Those are probably still laughing.

RE Paula not threatening to sue: that falls into the same category as her not getting paced in London and her racing at 29 C in October 2003. This is no surprise.

casual obsever wrote:

ad 3: Just as ludicrous. High Hct caused by altitude comes along with lower RET-%, and hers in 2012 were over 50% higher than in 2003.

Both, however, can be explained with using EPO at altitude.

Again, the IAAF didn't discuss this.

I don't know enough about the science to know about this. She claims she was injured when she was in Kenya in 2012 and therefore wasn't running, and therefore not crushing red blood cells, hence the high Hct. What would that likely do to the RET-%?

(I wonder how she proved she was injured in Kenya and then (one imagines, for the story to hold water) subsequently in the 10 days following when she came home, prior to the test? It's a long-time to be injured and not running for an elite athlete........)

Well, as far as I know, she didn't prove it. The IAAF likes to take her word.

There were also reports of her training during that time... but maybe she only trained 100 km instead of 160 km/week or so.

But yes, the less you train, the more normal your Hct. Normal, not raised.

Interesting is also a comparison of her altitude stays. In 2003 after altitude, she had [Hb] = 12 g/dL (and said in her bio that that low number was likely coming from her illness), in 2005 it was in the low 13s (comp. time), and in 2012, in the low 16s (base training). RET-% were at, or just below, 0.5% in 2003/05, and 0.8% in 2012. It would be quite instructive to see the other years, but now Paula went from demanding transparency from everyone to an advocate of no transparency.

From a legal point of view, the 2003 and 2005 values led to her being target tested (and thereafter, she went from a 2:15/2:17 runner to a 2:23 at best) because of the old rules, and in 2012 under the current set of rules, she narrowly escaped a ban because the expert panel decision was not unanimous. At least, that's the official version.

Interestingly the very same IAAF paper describes how they target tested Ramzi for two years without finding anything, until they finally caught him with CERA.

Wannabe statistician wrote:

Tucker's discussion of statistics alarms the statistician in me: "When the probability that a hematological value is not due to doping is about 1 in 1000, and only then with multiple flags and data points." I don't know how the biological passport works, and I mean, 1 in 1000 *sounds* pretty stringent, but if there are dozens or hundreds of tests per athlete (dozens of tests times a handful of markers) and thousands of athletes in the system then false positives are going to happen all the time. Tucker should know this.

No, you are mis-representing the probabililty. There are NOT dozens/hundreds of thousands of tests that give a hematological value which is 0.1% likely not to be due to doping.

The italicized part is what is claimed (and the 0.1%-barrier comes from a complementary construction, involving Bayesian implications about what "abnormal" is), while you are representing it as being that 1 in a 1000 tests are in the 0.1% percentile of OFF-scores.

The correct notion is that there are approximately tens/hundreds of "flags" a year that are over this 0.1% barrier, which again might lead to some Bonferroni corrections (if the same athlete multiply-fails), but in any case it's not what you are saying.

Well, here's the thing, the document says it's an ATPF if the result is outside of the 99th or 99.9th percentile for the expectations of THAT individual. 1/100 or 1/1000 may be different for me and for you, but it's still a probability of 1/100 or 1/1000 in either case! If you do a hundred tests, you're likely to get a 1/100 reading. Over the number of drug tests (Runner's World says about a dozen per year for top athletes) and number of statistical tests per drug test (markers, longitudinal tests, all the things you're looking at) and number of athletes being tested, there are going to be tons of false positives at the 1/100 or even 1/1000 level. Maybe the document is just worded imprecisely, perhaps they do make a correction for the total number of tests an athlete has taken over their life, I dunno

You are making the same mistake as others.

The condition Prob(person-is-doping | given OFF-scores) >= 99% is NOT a 1% occurrence.

Your condition is more like Prob(OFF-score is outside 99% range | given OFF-scores),

which tautologically is a 1% occurrence.

casual obsever wrote:

Well, as far as I know, she didn't prove it. The IAAF likes to take her word.

There were also reports of her training during that time... but maybe she only trained 100 km instead of 160 km/week or so.

Thanks for the reply.

What reports would those be - any source?

Here's what it says:

[quote]The Adaptive Model predicts for an individual an expected range within which a series of Marker values falls assuming a normal physiological condition. Outliers correspond to those values outside of the 99%-range, from a lower limit corresponding to the 0.5th percentile to an upper limit corresponding to the 99.5th percentile (1:100 chance or less that this result is due to normal physiological variation). A specificity of 99% is used to identify both haematological and steroidal ATPFs.[\quote]

They even use the word "specificity". They're telling us the false positive rate. Specificity = true negatives / (true negatives + false positives). If you measure 99 true negatives and 1 false positive then your specificity is 99/(99+1)=99/100 = 99%. Am I messing this up?

Can you explain this more thoroughly? The document says there only needs to be one test value that falls outside of the expected range:

This seems to disagree with your last paragraph, no? It's the results of the last test, considering previous tests. Regarding the Bonferroni correction, there's also this:

So how many statistical tests are there? For blood there is HGB and OFFS (2), plus the longitudinal profiles of HGB and OFFS (2+2=4), then for steroids there are 5 more (4+5=9) and their longitudinal profiles (9+5=14), so that's 14 statistical tests altogether that they're looking to fall within expected intra-individual ranges. I think it's fair to include steroidal tests because if Paula had failed one of those we'd be talking about it too. They're looking at 7 of those to fall within the 1:100 level and 7 to fall within the 1:1000?

If there are 2 blood tests per year and the athlete competes for 10 years, then that's 20 blood tests, times 14 statistical tests per blood tests, which makes 280 statistical tests. No?

I don't want it to sound like I'm fiercely defending the athletes by the way, like I said before I think they're all doping. I'm just reading something and confused by what I'm reading. I already regret writing so much about it because I'm sure someone is going to explain why I'm wrong and then I'll just be like... "oh" and the rest of the nonsense posts are going to stay on the board.

This is what the IAAF released after British Parliament implicated her .... "The IAAF ; Radcliffe was the victim of untrue allegations and was “hounded remorselessly in the media” after blood samples collected in Portugal in 2003, Finland in 2005 and Monte Carlo in 2012 were described as suspicious"

This was after a statement there was no evidence against , so they walked it back a bit.

DHT123 wrote:

What reports would those be - any source?

Here she says she trained well, twice a day during that altitude stay in Iten:

"Being in Iten really improved my confidence. I only expected to be running once a day and I ran twice, comfortably. I pushed hard and it worked out, and that gave me a huge boost. The trails are full of holes and lumps and bumps, and I was running hard on them. I gave it every chance for something to go wrong, and nothing did. If I survived that, I can survive anything "