So how much of a pinch of salt do we need to take it with? Oh that's right, we can't accurately adjust for threats to validity - that's what makes them threats to validity. So we don't actually know with a huge degree of certainty that there was any difference between the frames. Just like we don't have any confidence that a drug works without a double-blind placebo controlled trial, we don't have any confidence that two frames are different using your methodology when the rider knows what he's riding and has the ability, intentionally or otherwise, to adjust his position.

no, i said it's possible, which means it also might not.

so in the real world you mean to tell me that Fabian Cancellara loses 2 minutes on his 40k tt if he simply swaps to the P2K frame? Ha, that's laughable!

Maybe it's not hard to imagine, but it's hard to make it fit the observations. The P2K CdA estimated during this test run is consistent with the value he has estimated both during other tests and during races, so he would've had to have been sitting up during every one of those tests, too, in order to get the level of repeatability he's shown.

Hmmm. Now you're going off the rails. Desperation can do that. Calm down.

For example more reality........with a ~3 minute advantage, I'd expect some more (or maybe at least even 1?) P3C's !

Results - Stage 13 (Albi - Albi)
1. Alexandre Vinokourov (Kz), ASTANA, 1:06:34
2. Cadel Evans (Aus), PREDICTOR - LOTTO, 1:14
3. Andréas KlÖden (G), ASTANA, 1:39
4. Andrey Kashechkin (Kz), ASTANA, 1:44
5. Bradley Wiggins (GB), COFIDIS CREDIT PAR TELEPHONE, 2:14
6. Yaroslav Popovych (Ukr), DISCOVERY CHANNEL TEAM, 2:16
7. Alberto Contador (Sp), DISCOVERY CHANNEL TEAM, 2:18
8. Sylvain Chavanel (F), COFIDIS CREDIT PAR TELEPHONE, 2:38
9. Levi Leipheimer (USA), DISCOVERY CHANNEL TEAM, 2:39
10. Mikel Astarloza (Sp), EUSKALTEL - EUSKADI, 2:42
11. Michael Rasmussen (Dk), RABOBANK, 2:55
12. Vladimir Gusev (Rus), DISCOVERY CHANNEL TEAM, 2:56
13. Leif Hoste (B), PREDICTOR - LOTTO, 2:56
14. Linus Gerdemann (G), T-MOBILE TEAM, 3:09
15. Juan Manuel Manuel (Sp), QUICK STEP - INNERGETIC, 3:12
16. Juan Jose Cobo Acebo (Sp), SAUNIER DUVAL - PRODIR, 3:13
17. Vladimir Karpets (Rus), CAISSE D’EPARGNE, 3:17
18. Kim Kirchen (Lux), T-MOBILE TEAM, 3:18
19. Oscar Pereiro Sio (Sp), CAISSE D’EPARGNE, 3:23
20. David Millar (GB), SAUNIER DUVAL - PRODIR, 3:27

Check the prologue and the final TT, same thing, 1 or no P3C's........with that huge advantage, sorry but this theory isn't adding up.....

Likewise, if P3C is 2 min faster than P2K, then P2K is probably a good 3 min faster than round tube. And a round tube bike with the same geometry " ain't no " 5 minutes slower than a P3C no matter how you look at it!

I'm calm enough to address your assertions with logic and accurate critique of your experimental method. Care to do the same regarding my comments above?

Dude, I'm not the one who is attributing your criticisms of my argument to "threats to validity." That's how you're viewing my responses. That's desperate, and paranoid. We've already seen Tom's repeatability across trials, and you can inspect for yourself his repeatability within the trials by examining the P2K and P3C plots in the very first post in this thread. If his position were changing, the profiles would be variable. Take a look. How variable are they? The P2K profiles are rock solid. You can see small amounts of shape difference across P3C laps but nowhere near .023 m^2 worth -- more like .002 m^2 worth. That's what I'd call a pinch of salt.

BTW, if the laps are done at different speeds (as they were here) then wind also distorts the profiles. That's how we know that whatever wind there was must have been very, very small.

Dude, I'm not the one who is attributing your criticisms of my argument to "threats to validity." That's how you're viewing my responses.[/reply] Could you fill me in on what this sentence means? The lack of blinding in your observational study threatens the validity of your conclusions. That's what I mean by "threats to validity".

why?

if the power output is the same, conditions are the same, why not?

what experiment will it take to prove this out to you?

PI suggestion that if perhaps if tom found a 10k course that was dead flat, out and back, close to a weather station that would show that conditions remained the same (or dammit Tom you could get a kestrel) and made two runs at the same power (Pave~=Pnorm or VI=1.0) back to back and the difference was as predicted by the field testing (~30s) difference...

would you and FD believe?

would you believe that FC would loose 2 minutes in a 40k tt by switching to a p2k?

if not, then would the two of you PLEASE describe a proof that would satisfy you?

and barring that quit yer bitching?

because the "intellectual" part of the discussion is rapidly degrading....

g

---

greg
www.wattagetraining.com

if not, then would the two of you PLEASE describe a proof that would satisfy you?

and barring that quit yer bitching?

because the "intellectual" part of the discussion is rapidly degrading....

g[/reply]
that's the argument, that there really isn't a "Proof" that would be totally accurate across the board. no doubt that it's and interesting study, but if the goal is to say a P3C is definitely 2min faster as a rule of thumb, then i think you're way off base with that type of assertion.

it's not that "intellectual" if you try to turn a one study opinion into fact, without considering all potential sides of the story, the likelyhood of flaws, and as well as trying to dis-prove the theory. now if we start talking multiple people doing the same test and get similar/same results, then we might be on to something.

if the claim is simply "it's faster for Tom A" then i'm totally on board with you.

I have already stated what proof would satisfy me. That is a statistically significant analysis looking at the time improvements of a large group of people of known ability who change bikes to a P3C or the time slowing if they change from a P3C to another bike.

the current test done by Tom suggests such an explanation, but, in view of the largeness of the difference, the lack of the real world (racing) correlation suggests to me another explanation for the difference is at work.

---

--------------
Frank,
An original Ironman and the Inventor of PowerCranks

I don't think tom ever made a claim beyond that, BUT...

I would say the the refinded claim should read:

"
The p3c is significantly faster for tom vs. the generally acceptly fast p2k. And there does appear to be a strong correlation to the data of at least 2 other riders that suggest the difference is rider independant"

G

---

greg
www.wattagetraining.com

The Trimble had a boxed-off area behind the "seatstays" which IMO would result in a large low-pressure zone there, and resulting high drag. I'd be willing to bet that the vortices spinning off the rider's legs interacted with that zone substantially. I remember looking at that bike (1990 Nats in Albany or 1991 at Tooele IIRC) and thinking "there's no way this thing can be aero".

FWIW, Eric

One more time: are you going to provide some evidence for your assertions?

Really, can't you see that this is getting tiresome? You trot out one unsupported statement after another, but you never provide any reasoning, any evidence, or any logic - just your unsubstantiated beliefs. If you want to argue about religion, fine, but maybe there's a better place for that.

How come you don't argue for the same rigor in the promotion of your product Frank? You seem awfully quick to promote the "successes" of n=1 experiments with absolutely no control of whether it was simply training by itself or training with PCs. A bit hypocritical isn't it? Now if you are arguing for better control across the board...

I wish those Trimbles were gone. Some guy was in a crit last week with one. Drove me nuts.

Even if you have that low pressure in the back, in a yaw condition, I would expect you get a benefit similar to a disc wheel.

"Likewise, if P3C is 2 min faster than P2K, then P2K is probably a good 3 min faster than round tube. And a round tube bike with the same geometry " ain't no " 5 minutes slower than a P3C no matter how you look at it!"

Yet again: are you going to provide some evidence for your assertions? Your first sentence above is a marvelously contorted logical absurdity. The second one I can accept as your opinion, sure, but again, it is your unsubstantiated belief and why on Earth should we care?

I really don't intend to pick on you ; there's at least one other poster on this thread who all my comments would apply to equally well. However, he has proven to be incorrigible and I was hoping you weren't. Having asked several times now for you to support your assertions, and having gotten only more of the same, I'm bowing out.

Could you fill me in on what this sentence means? The lack of blinding in your observational study threatens the validity of your conclusions. That's what I mean by "threats to validity".[/reply]You're asking me to fill you in on a sentence you wrote? Yikes.

Anyway, perhaps you will be able to see why you're coming off as desperate when I explain why I'm not threatened at all: this isn't my study, observational or otherwise. You haven't made any argument except to point out the potential for procedural bias. The potential bias you're hanging you hat on is of unknown size but you're saying it invalidates any claim to difference.

I've been pointing out a difficulty in your argument: that the profiles in this test in conjunction with results from Tom's earlier tests put bounds on the size of a putative procedural bias. Those bounds are much smaller than the size of the effect estimated here.

So don't argue "the potential for bias is so huge it invalidates the results." You have to argue "the bias is so huge it invalidates the results." And, if you're going to stick to that argument, try to explain how the bias could be that huge and yetl be undetected by the data before us.

You're asking me to fill you in on a sentence you wrote? Yikes.

Anyway, perhaps you will be able to see why you're coming off as desperate when I explain why I'm not threatened at all: this isn't my study, observational or otherwise. You haven't made any argument except to point out the potential for procedural bias. The potential bias you're hanging you hat on is of unknown size but you're saying it invalidates any claim to difference.

I've been pointing out a difficulty in your argument: that the profiles in this test in conjunction with results from Tom's earlier tests put bounds on the size of a putative procedural bias. Those bounds are much smaller than the size of the effect estimated here.

So don't argue "the potential for bias is so huge it invalidates the results." You have to argue "the bias is so huge it invalidates the results." And, if you're going to stick to that argument, try to explain how the bias could be that huge and yetl be undetected by the data before us.[/reply] Maybe I could accept this result if the pro tour teams were paying Cervelo to ride their bikes, instead of the other way around.

I don't know the actual size of the bias. I know it to be potentially large. If we can all agree that there is a potential bias and that no one knows how large it is then maybe we can all agree that it is unreasonable to argue that the totality of this result represents aerodynamic improvement attributable to the bike itself.

---

--------------
Frank,
An original Ironman and the Inventor of PowerCranks

You're asking me to fill you in on a sentence you wrote? Yikes.

Anyway, perhaps you will be able to see why you're coming off as desperate when I explain why I'm not threatened at all: this isn't my study, observational or otherwise. You haven't made any argument except to point out the potential for procedural bias. The potential bias you're hanging you hat on is of unknown size but you're saying it invalidates any claim to difference.

I've been pointing out a difficulty in your argument: that the profiles in this test in conjunction with results from Tom's earlier tests put bounds on the size of a putative procedural bias. Those bounds are much smaller than the size of the effect estimated here.

So don't argue "the potential for bias is so huge it invalidates the results." You have to argue "the bias is so huge it invalidates the results." And, if you're going to stick to that argument, try to explain how the bias could be that huge and yetl be undetected by the data before us.[/reply] I still don't understand what your sentence meant. Whether it's the problematic grammar of your sentence or your misunderstanding of the term "threats to validity" I'm not sure, but it's beside the point.

I apologise for attributing this study to you. It's Tom's study, but my criticisms are of the study design which, if I understand right, is attributable to you. I actually think it's a good study design and, with the evidence presented, I'd be pretty confident that the P3C is a faster frame for Tom than the P2K. I don't think it's valid, though, based on the evidence presented, to suggest that you can estimate of the magnitude of difference attributable solely to the frame with much precision. I think a "2 seconds per km" or similar rule of thumb isn't well supported by the evidence presented.

Since you brought up reality, you should probably keep in mind that these are all different riders with different strengths, different fatique levels, and most importantly DIFFERENT GOALS (among many other things). These things will certainly bias the outcome. You see, aerodynamics isn't the only thing to consider when you are talking about a single stage in the pro tour, even if it's a TT.

---

---

Erik
Strava

I really wanted to thanks sib1 and donm for what they did in this thread. Through their clever questions and remarks, the truth surfaced at last, and now we know that this kind of testing, even if interesting, has too many faults and doesn't relate at all to the real world. Thanks guys.

that's just like you. Except for the interesting part.

John

---

---

Top notch coaching: Francois and Accelerate3 | Follow on Twitter: LifetimeAthlete |

[/reply]
One more time: are you going to provide some evidence for your assertions?

Really, can't you see that this is getting tiresome? You trot out one unsupported statement after another, but you never provide any reasoning, any evidence, or any logic - just your unsubstantiated beliefs. If you want to argue about religion, fine, but maybe there's a better place for that.[/reply]
no, like i said before, think about it in the real world with real pro's. i'm asking you to think outside the scientific box for 2 seconds and see if that would hold up in reality. if you want to think that FC would lose 2 minutes by swapping a great frame for a good frame, then that's your problem. i think the real world pro results speak for themselves. figured up the ratio of guys on Cervelo's vs winners on Cervelo's, and it isn't too tough to see it doesn't add up!

Since you brought up reality, you should probably keep in mind that these are all different riders with different strengths, different fatique levels, and most importantly DIFFERENT GOALS (among many other things). These things will certainly bias the outcome. You see, aerodynamics isn't the only thing to consider when you are talking about a single stage in the pro tour, even if it's a TT.[/reply] yes, which is why i said this TT stage (middle), the last TT stage (last), the prologue (first), Ironman Hawaii, on and on and on......do the math here, where is this huge ratio of Cervelo winners? with a 2 minute head start they should be winning more, and a 10 minute head start at Hawaii they should be winning more, hmmmmm