In Reply To:
In Reply To:
In Reply To:
In Reply To:
In Reply To:
For instance, a current PC study going on right now the person doing the testing is blinded at to what person is doing what. The participants are not blinded, obviously, but the tester/evaluator is.
There's really no room for judgment calls or evaluator bias on the analytical side: Tom is using a spreadsheet assembled by Alex Simmons using formulas that are open and inspectable. And it would be extremely difficult for someone to manipulate the data in a way that would 1) produce an effect of the size he's showing, 2) make the profiles look the same, and 3) be undetectable. Do you recall the last time I made a statement similar to that?
But, that is not the only potential source of error. How about the accuracy or precision of the power meter?
Tom used the same power meter so even if there was an error in the meter it wouldn't explain the difference between the estimated CdAs.
It could, if the precision of the meter is not perfect. Since no measurement device is perfect, we can presume some error could come from this. Only question is, how much? What is the precision of the meter he used?
But, more than this. He is attributing all of the change to the bicycle. This is ludicrous in view of the fact we are talking about humans on a bicycle. It is simply impossible for them to be exactly the same on these different trials. The overall result could be entirely accurate but to attribute the change entirely to the difference in bicycle frames seems a bit of a stretch.
again please RTFM over and over until you can recite it by chapter and verse. The quiz will be on Monday .. But to make one point, each trial is effectively multiple trials (laps) and us visual monkeys are darned good at picking out patterns that match (or do not!). Something about the jungle I suppose ... So while I did query the magnitude of the delta, the noise ain't nearly as bad you suggest. Repeatability is ~0.001 m^2 or ~1W. Yes, that implies the PM is repeatable to around 1W. I have about five million step tests vs. my PT Pro and CT and two million with my SRM and CT from 150 to 450W and they show the same thing. Repeatability is considerably better than the stated accuracy. I have no problem with that ... Anyhow, carry on ...
I will accept your assessment of the accuracy of this result. Do you believe it is correct to assign all the change seen to the bicycle?
as Tom's previous testing (many different setups that had nothing to do with this particular arrangement), has shown a repeatability of ~0.001, it's is quite reasonabe to state his results for this test as showing a delta CdA of 0.023 +/- 0.001-0.002 m2.
So, no, I do not believe 100.00000% of the measured changes were due to changing bikes. I believe at least 95% were and the other 5% due to various noise factors amongst which changes in position are included.
Yesterday I was analyzing a sizeable data set for someone. This was in a different country, different conditons, different monkey and whilst riding in unfamiliar surroundings. The repeatability of the results taking the baseline and repeating that test at the end: 0.001 m2 !!!
It's either a global conspiracy or structured field testing really can produce results that consistent. I figure #2.