Data (was PowerCranks Study)

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

Quoted message said:

Frank Day said:

It seems to me, from a scientific point of view (and from the point of view of the
non-believers) that the first thing that needs to be proven is this question: Is there any worth
to these things over what can be done without them.

Once that is accepted then several questions arise such as how much potential benefit is there,
what is the optimum way to achieve that benefit and what is the mechanism or mechanisms of the
benefit.

Exactly. A better experiment would include other methods of pedal stroke

improvement.

Quoted message said:

For example, a full-factorial between PC, one-legged pedaling, big gear

climbing,

Quoted message said:

downhill spin drills, and placebo approaches. Get, say, 10000 cyclists,

and split

Quoted message said:

'em up.

I wondered about basically the same thing: how much of the reported improvement in gross efficiency
in the treatment group was due to use of PowerCranks per se, and how much was due to the fact that
they spent 6 weeks thinking about their pedaling? I don't think you'd necessarily have to compare
every conceivable method of improving pedaling "skill" to have a valid comparison, but hopefully the
non-treatment group was exposed to some sort of "attention control", and didn't just keep on
training as usual. If they did, then interpreting the results is quite difficult.

Two other questions (for anybody who might know the answer):

1) what *did* happen to net or delta efficiency? If there were no improvements there, then
something's definitely amiss.

2) who funded the study?

Andy Coggan

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

Quoted message said:

Hurdler,

I say "n number" because I want the laypeople in this group to understand why it's wrong.

Uh-huh.

I am aware what the

Quoted message said:

variable "n" signifies in statistical computations. Let's not hit below the belt.

You and Chris both say you can get "good" results with a 12 subject study. The problem with your
response is you're using the word "good." No scientists uses that word to describe a data set
because it's a relative adjective.

I don't think I used the word 'good.' But valid, usable results are possible in some cases with n =
12. Whether that is the case when you have 6 people ride normal bikes and 6 ride bikes with a broken
bottom bracket, I'm not sure.

Quoted message said:

That bike is "good." What does that really mean?

That it's not a Bianchi?

Quoted message said:

What you are also NOT saying is that you can get MUCH BETTER results using 200 subjects.

That's actually exactly what I said. You can get better still if you use
1,000,000,000,000,000,000,000,000 subjects.

And the difference

Quoted message said:

between good results and MUCH BETTER results I submit to you will exceed the 2% efficiency claimed
in this study by the Power Cranks study director.

Why don't you tell me that you can also get "good results" with 8 subjects, or 4 sinjects? How
about 2 subjects?

Come back on this newsgroup and tell me exactly how you would rate this same study if it were done
using 2 subjects, 4 subjects, 6 subjects, 8 subjects, 10 subjects, and 200 subjects, etc.

After about n = 10 I'd say the studies were 'good.'

Quoted message said:

You see where I'm going with this argument, don't you?

Uh, no.

I will prove to you using your own keyboard that you will

Quoted message said:

agree with me that as the "n" increases, the results get "better." At some point, the n issue
becomes neglgible
(i.e. when 200 subjects are used). But using 12 subjects is not good technique.

Number of subjects isn't really part of 'technique.'

Quoted message said:

Studies should be conservative, not risk-taking ventures. If you want me to believe in the
results, use 200

Quoted message said:

I threw the number 200 out there because...

because you don't really know what you're talking about.

that's a number that isn't going to be attacked easily for an "n"

Quoted message said:

problem. When you intentionally design a study that uses 12 subjects, you are telling me you are
the kind of person who folds his homework up in his back pocket and does the minimal amount of
work possible to get by. You're a "C" student and when you don't get an A (I would get an A
because I used 200 subjects in my experiement model), you call the teacher a bunch of names and
say it's "good enough."

Sorry, but the results of a 12 person study don't impress me like the results of a 200 person
study would. No scientist will refute that argument.

Not even you.

Racer X

Spritle -

Solve for blank:

n X _________ = statistical significance.

-Kabala

speedracer.comchar.htmOpen ↗

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

Quoted message said:

RACER X said:

Trust me, I know statistics. I've done graduatel level work with DNA allele frequency statistics
- that calculus stats.

Oh dear.

Horrifying.

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

Quoted message said:

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

Quoted message said:

Frank Day said:

It seems to me, from a scientific point of view (and from the

point of

Quoted message said:

Quoted message said:

Quoted message said:

view of the non-believers) that the first thing that needs to be proven is this question: Is
there any worth to these things over

what

Quoted message said:

Quoted message said:

Quoted message said:

can be done without them.

Once that is accepted then several questions arise such as how

much

Quoted message said:

Quoted message said:

Quoted message said:

potential benefit is there, what is the optimum way to achieve

that

Quoted message said:

Quoted message said:

Quoted message said:

benefit and what is the mechanism or mechanisms of the benefit.

Exactly. A better experiment would include other methods of pedal

stroke

Quoted message said:

improvement.

Quoted message said:

For example, a full-factorial between PC, one-legged pedaling, big

gear

Quoted message said:

climbing,

Quoted message said:

downhill spin drills, and placebo approaches. Get, say, 10000

cyclists,

Quoted message said:

and split

Quoted message said:

'em up.

I wondered about basically the same thing: how much of the reported improvement in gross
efficiency in the treatment group was due to use

of

Quoted message said:

PowerCranks per se, and how much was due to the fact that they spent 6

weeks

Quoted message said:

thinking about their pedaling? I don't think you'd necessarily have to compare every conceivable
method of improving pedaling "skill" to have

a

Quoted message said:

valid comparison, but hopefully the non-treatment group was exposed to

some

Quoted message said:

sort of "attention control", and didn't just keep on training as

usual. If

Quoted message said:

they did, then interpreting the results is quite difficult.

Two other questions (for anybody who might know the answer):

1) what *did* happen to net or delta efficiency? If there were no improvements there, then
something's definitely amiss.

Neither of these were mentioned in the study.

Quoted message said:

2) who funded the study?

Frank reported previously ..."No one paid this group to do this study (although I did provide one of
the authors a pair of cranks to use for the study)"

To be fair to the authors, further discussion of the details should wait until the study is
published. I was a little premature in offering to send out copies. The least I can hope for is
this study will cause others to look more closely and subsequent studies will address any
shortcomings of this one.

Phil Holman

"Andy Coggan" <[email hidden]> wrote in message news:<EKwhb.4684$av5.35

Quoted message said:

Two other questions (for anybody who might know the answer):

1) what *did* happen to net or delta efficiency? If there were no improvements there, then
something's definitely amiss.

2) who funded the study?

Andy Coggan

Most of the research into pedaling (funded or unfunded) has been a waste of time and money because
for as long as the "experts" focus all their research on the natural ( round or stomping)styles,
all they are doing is trying to " dig their way out of a hole ". Linear pedaling (Anquetil's) can
biomechanically eliminate the dead spot area (11 to 1 o'clock) by making it part of the main power
stroke, enable a rider to biomechanically combine the upper body power of arm/hand cranks with the
lower body power of leg cranks and completely eliminate the root cause (natural pedaling) of all
cycling related lower back pain. You will not get a more powerful, smoother or safer pedaling
technique and all can be done by simply making more sensible use of your muscles. Powercranks
advantage comes from forcing a rider to unweight the pedal of the idling leg as this has a double
effect of increasing the effectiveness of the power application to the pedal when the round or
stomping style is used. For that reason I would see them more as training equipment for young
riders or those who do not have the perfect unweighting technique.

n crowley said:

Linear pedaling

Is that anything like nlroei.nlFotoboek display 23674.htmlOpen ↗ ?

Ewoud Dronkert said:

n crowley said:

Linear pedaling

Is that anything like nlroei.nlFotoboek display 23674.htmlOpen ↗ ?

This looks like it uses linear foot propulsion: kickbike.nlkickbike.nlOpen ↗

Robert Chung said:

This looks like it uses linear foot propulsion: kickbike.nlkickbike.nlOpen ↗

Aaaargh! Have these people no shame! To include a valve nut and -cap on the racer model, jeez.

Based on your background in statisitics, I am surprised by your insistance that a higher number of
participants means a more valid or acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a study is
valid or not is whether the statistical significance level reached .95. It doesn't matter if a study
has 8, 12, 200, 2,000, or a million participants, if the statisical significance is reported as
being at the .95 level then each of these studies are equally valid. They would each have a 1 in 20
chance that the findings were due to chance and a 19 in 20 chance it is not.

It is only necessary to have "enough" participants to reach statistical significance to have a
"valid" study. Anything more is a waste of resources.

Frank

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

Quoted message said:

Hurdler,

I say "n number" because I want the laypeople in this group to understand why it's wrong. I am
aware what the variable "n" signifies in statistical computations. Let's not hit below the belt.

You and Chris both say you can get "good" results with a 12 subject study. The problem with your
response is you're using the word "good." No scientists uses that word to describe a data set
because it's a relative adjective.

That bike is "good." What does that really mean?

What you are also NOT saying is that you can get MUCH BETTER results using 200 subjects. And the
difference between good results and MUCH BETTER results I submit to you will exceed the 2%
efficiency claimed in this study by the Power Cranks study director.

Why don't you tell me that you can also get "good results" with 8 subjects, or 4 sinjects? How
about 2 subjects?

Come back on this newsgroup and tell me exactly how you would rate this same study if it were done
using 2 subjects, 4 subjects, 6 subjects, 8 subjects, 10 subjects, and 200 subjects, etc.

You see where I'm going with this argument, don't you? I will prove to you using your own keyboard
that you will agree with me that as the "n" increases, the results get "better." At some point,
the n issue becomes neglgible
(i.e. when 200 subjects are used). But using 12 subjects is not good technique.

Studies should be conservative, not risk-taking ventures. If you want me to believe in the
results, use 200

Quoted message said:

I threw the number 200 out there because that's a number that isn't going to be attacked easily
for an "n" problem. When you intentionally design a study that uses 12 subjects, you are telling
me you are the kind of person who folds his homework up in his back pocket and does the minimal
amount of work possible to get by. You're a "C" student and when you don't get an A (I would get
an A because I used 200 subjects in my experiement model), you call the teacher a bunch of names
and say it's "good enough."

Sorry, but the results of a 12 person study don't impress me like the results of a 200 person
study would. No scientist will refute that argument.

Not even you.

Racer X

Ronaldo Jeremiah said:

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

Quoted message said:

Like I said Frank, using an "n" number of 12 is already such a significant fault of the study
design, that all you people touting its conclusions are just reinforcing your total lack of
understanding of items like standard deviation, statistics in general, and imprecision of
measuring equipment.

All your other discussions are moot because you'll NEVER convince any real scientist that
using 12 subjects can generate meaningful statistics of any kind.

Like I said, it's a fatal flaw and any further discussions of this study that attempt to gloss
over the "n" number problem cannot pass the laugh test.

Racer "Johnny Cochran" X

RACER X:

You've been spending too much time watching Speed Racer (Trixie's hot, ain't?). If you spent
less time watching cartoons and more time reading about science and statistics, you'd know that
'n' *means* 'number,' and you'd stop looking like a fool by saying 'n number.' You'd also know
that having n = 12 is not a 'fatal flaw.' Sure, a higher n would be better - it always is. But
if they achieved statistical significance with n = 12, that may actually indicate a strong
effect. But you don't know what I'm talking about, so I'll just leave it at that.

When your understanding is incomplete, it makes the correct conclusion impossible.

It's like that episode of Speed Racer with the Mammoth Car. All along, Speed and his friends
thought the car was being used to *smuggle* the gold. In the end, it turned out that the Mammoth
Car WAS the gold. WOW. That cartoon kicked ass.

You're really doing a disservice to the name of Zen-Master RACER X. Cease and desist.

-RJ, n = 1.

Frank Day said:

Based on your background in statisitics, I am surprised by your insistance that a higher number of
participants means a more valid or acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a study
is valid or not is whether the statistical significance level reached .95.

Oh dear.

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

Quoted message said:

Ewoud Dronkert said:

n crowley said:

Linear pedaling

Is that anything like nlroei.nlFotoboek display 23674.htmlOpen ↗ ?

This looks like it uses linear foot propulsion: kickbike.nlkickbike.nlOpen ↗

Linear pedaling is a technique which enables a seated rider to reproduce the same power as an "out
of the saddle" rider by using most of the muscles in the same way and with the advantage of arm
power and most of the body weight when required. This is done by making greater use of the hips and
calf muscles. When riding out of the saddle, the muscles above the knee can produce the same power
with less effort than that of a seated rider. Eliminating the dead spot area enables a rider to turn
a higher gear than that of rider who idles through the dead spot area.

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

Quoted message said:

Frank Day said:

Based on your background in statisitics, I am surprised by your insistance that a higher number
of participants means a more valid

or

Quoted message said:

Quoted message said:

acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a study
is valid or not is whether the statistical significance level reached .95.

Oh dear.

mathworld.wolfram.comProbabilityandStatistics.htmlOpen ↗

Phil Holman

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

Quoted message said:

Frank Day said:

Based on your background in statisitics, I am surprised by your insistance that a higher number
of participants means a more valid or acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a study
is valid or not is whether the statistical significance level reached .95.

Oh dear.

Indeed.

-RJ

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

Quoted message said:

Frank Day said:

Based on your background in statisitics, I am surprised by your insistance that a higher number
of participants means a more valid or acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a study
is valid or not is whether the statistical significance level reached .95.

Oh dear.

apa.orgpsp7761121.htmlOpen ↗

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

Quoted message said:

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

Quoted message said:

Frank Day said:

Based on your background in statisitics, I am surprised by your insistance that a higher
number of participants means a more valid

or

Quoted message said:

Quoted message said:

acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a
study is valid or not is whether the statistical significance level reached .95.

Oh dear.

mathworld.wolfram.comProbabilityandStatistics.htmlOpen ↗

Phil Holman

Um, Phil? In case you didn't know it, Robert's field of research requires that he have more than a
passing knowledge of statistics. As well, he can be one of the more oblique posters...IOW, his "oh
dear" should not be taken as commentary on Frank Day's comment regarding statistics per se.

Andy Coggan

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

Quoted message said:

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

Quoted message said:

Frank Day said:

Based on your background in statisitics, I am surprised by your insistance that a higher
number of participants means a more valid

or

Quoted message said:

Quoted message said:

acceptable study and a lesser number is somehow inferior.

Assuming the design and other elements are done acceptably, the determinant of a whether a
study is valid or not is whether the statistical significance level reached .95.

Oh dear.

mathworld.wolfram.comProbabilityandStatistics.htmlOpen ↗

Phil Holman

Hey, as long as we are talking stats -

What are the odds that Imelda Hincapie wins a race that's held on a weekend during the 2004 season?

-RJ, just a stone's throw from Mathworld...

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

Quoted message said:

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

Quoted message said:

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

Quoted message said:

Frank Day wrote:
> Based on your background in statisitics, I am surprised by your insistance that a higher
> number of participants means a more

valid

Quoted message said:

Quoted message said:

or

Quoted message said:

> acceptable study and a lesser number is somehow inferior.
>
> Assuming the design and other elements are done acceptably, the determinant of a whether a
> study is valid or not is whether the statistical significance level reached .95.

Oh dear.

mathworld.wolfram.comProbabilityandStatistics.htmlOpen ↗

Phil Holman

Um, Phil? In case you didn't know it, Robert's field of research

requires

Quoted message said:

that he have more than a passing knowledge of statistics. As well, he

can be

Quoted message said:

one of the more oblique posters...IOW, his "oh dear" should not be

taken as

Quoted message said:

commentary on Frank Day's comment regarding statistics per se.

Andy Coggan

Ditto on me posting the link. I'm aware of Robert's background (he sent me the full article on an
abstract I was interested in) and his writing style. OTOH in case there are any further misgivings,
it's a very handy reference link.

Phil Holman

I can assure you that there are plenty of people out there (many of them at this site) who know more
statistics than I do. I did find his remark somewhat obscure and difficult to figure since he made
the exact same remark earlier on a completely different remark. After reading that comment I did
reread my posting and thought my remarks were possibly susceptible to misinterpretation as to my
real meaning.

Despite my ignorance of the nuances of statistics, I did pay enough attention and absorb enough
(I think) that I learned that those significance probability numbers were developed just so us
mere mortals can still read these studies with some ability to independently judge the worth of
the study.

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

Quoted message said:

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

Quoted message said:

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

Quoted message said:

Frank Day wrote:
> Based on your background in statisitics, I am surprised by your insistance that a higher
> number of participants means a more valid

or

Quoted message said:

Quoted message said:

> acceptable study and a lesser number is somehow inferior.
>
> Assuming the design and other elements are done acceptably, the determinant of a whether a
> study is valid or not is whether the statistical significance level reached .95.

Oh dear.

mathworld.wolfram.comProbabilityandStatistics.htmlOpen ↗

Phil Holman

Um, Phil? In case you didn't know it, Robert's field of research requires that he have more than a
passing knowledge of statistics. As well, he can be one of the more oblique posters...IOW, his "oh
dear" should not be taken as commentary on Frank Day's comment regarding statistics per se.

Andy Coggan

Andy Coggan said:

As well, he can be one of the more oblique posters...

Oblique? Moi?

My point of view is that statistical validity isn't like a lightswitch with on and off positions:
research findings aren't necessarily invalid if the significance level is 94% nor do they
necessarily become valid when its significance level reaches 96%. Research validity has many more
dimensions to it than statistical validity. There are *so many* ways studies can go bad even if a
p-value looks good.

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

Quoted message said:

Andy Coggan said:

As well, he can be one of the more oblique posters...

Oblique? Moi?

My point of view is that statistical validity isn't like a lightswitch with on and off positions:
research findings aren't necessarily invalid if the significance level is 94% nor do they
necessarily become valid when its significance level reaches 96%. Research validity has many more
dimensions to it than statistical validity. There are *so many* ways studies can go bad even if a
p-value looks good.

That's what I thought you meant (and I concur). Unfortunately, there are those (typically not
work-a-day scientists) that live and die by the P value.

Andy Coggan