New study shows leg flexion less efficient than extension.

[elapid]

Re: Re:

I would just like to point out that Dr. means teacher, at least that is what I was told in medical school. As I said, I see this way of "debating" as simply academic bullying.

Frank, you went to medical school with the dinosaurs and a lot has changed since then: doctor is a clinician or researcher (depending on the degree and whether it is honorary [medical doctor, veterinarian, dentist, etc] or research [PhD]), and a teacher is a teacher.

As for debating, you do not know the meaning of the word. You have your opinion/biases and you have hardly ever been swayed from your opinion/biases despite being proven wrong on many occasions. The only occasion that I can recall where you have ever conceded that you were wrong was a long, drawn out thread about whether pedalling was a reflex action or not, and even then you only half-conceded. So, as far as your post goes ... pot meet kettle.

 

[FrankDay]

Re: Re:

...
So, given the two examples I used how should the lay person reading this thread interpret those to data results. In one instance two groups are compared and the difference in power and efficiency change between the two groups reached the 0.125 and 0.25 level. What should that mean to the reader?
...

----------------------------------------------------
The p=0.125 means that If the results were due to chance, that there is a 12.5% probability that the results that were obtained would happen. Or conversely, there's an 87.5% probability that other (less meaningful) results would happen.

That is incorrect, at least as I interpret what you are saying. Again from Wikipedia article on statistical significance

The p-value is the probability of obtaining at least as extreme results given that the null hypothesis is true

A p of .125 means there is a 12.5% chance of getting the measured difference if one assumes the two groups are identical or an 87.5% chance the two groups are not identical such that the difference seen is not due entirely to chance.

What that means to the 'reader' varies ....
If the reader is looking for 'proof of validity', then 12.5% is not good enough.

That is by convention but reaching p<0.05 does not mean that there is no chance the difference seen is not due to chance.

If the reader is looking for an 'indication', then 12.5% might be adequate.

indeed.

 

[JayKosta]

Re: Re:

...
Do you have a link to such a definition? I found this "In inferential statistics the null hypothesis usually refers to a general statement or default position that there is no relationship between two measured phenomena, or no difference among groups
...

--------------------------
I'm confused about the idea of there being 'no difference amoung groups' - what does that mean? If the test results from different groups are the same (no differences among them), that says that the intervention that was being being tested didn't produce any observed effects that would result in a difference.

The part of the definition that is of most interest to our discussion is 'there is no relationship between two measured phenomena' . Which means that YES there can be differences, but they are the result of 'chance', and not due to some 'cause and effect' relationship.

Jay Kosta
Endwell NY USA

 

[FrankDay]

Re: Re:

...
Do you have a link to such a definition? I found this "In inferential statistics the null hypothesis usually refers to a general statement or default position that there is no relationship between two measured phenomena, or no difference among groups
...

--------------------------
I'm confused about the idea of there being 'no difference amoung groups' - what does that mean? If the test results from different groups are the same (no differences among them), that says that the intervention that was being being tested didn't produce any observed effects that would result in a difference.

The part of the definition that is of most interest to our discussion is 'there is no relationship between two measured phenomena' . Which means that YES there can be differences, but they are the result of 'chance', and not due to some 'cause and effect' relationship.

Jay Kosta
Endwell NY USA

The "no difference between groups" refers to the assumption of the null hypothesis. If there is no difference between groups actually measured (a virtual impossibility) then it is pretty clear the null hypothesis cannot be rejected. However, there is almost always a measured difference between groups so the purpose of the statistical analysis is to help the researcher make sense of the data as to the likelihood the measured difference between groups represents a real difference between the groups or not. It is simply a tool to help those trying to interpret data to figure out what may be important and what is not. The number per se means little without taking the entire context of the study into account. It is why Fergie claiming that a study that lasts 5 weeks of part-time use is hardly worth much in evaluating the PowerCranks claims that require 6-9 months of immersion training. Yet, he continues to think the fact these studies failed to reach the 0.05 threshold somehow proves something.

 

[JamesCun]

Re: Re:

And it seems that confusion about p-value is fairly widespread - see here for another discussion about its complexity.
http://blogs.plos.org/publichealth/2015/06/24/p-values/

Jay Kosta
Endwell NY USA

They are hugely misunderstood, misinterpreted and misrepresented, although this is true of many statistical values and methods (don't get me started on PCA). A lot of the time they're not even corrected! However it's extremely well known in science and there has been a steady push to improve on these matters as big data increases the requirements for statistics in areas of science where traditionally ratios and t-tests were the norm.

So someone on a cycling forum not really getting them is completely understandable, some of the smartest people I know don't really understand them. But they actually listen when people who do understand them tell them why they can't do what they are doing...

"They" don't listen to you because you don't tell them what they should be saying. I don't know what your academic background is but I suspect it is substantial. I would just like to point out that Dr. means teacher, at least that is what I was told in medical school. As I said, I see this way of "debating" as simply academic bullying. You point out where others are wrong (which of course makes you superior) but never say what is correct. I look forward to your eventually telling the group how one should properly interpret a p=0.125 between two groups showing a difference.

Frank, this has been explained to you over and over again. I'm quite surprised anyone is still willing to write anything with all of your bullying on this thread.

The interpretation is that the researchers didn't find a significant difference in their study. That is why they reported .125 and .25. If they had found a significant difference in the data, they would've reported that. What you are doing is misleading and constantly shifting your language to confuse the subject. You've repeatedly said that the p value is the probability of the null hypothesis being true. Others have repeatedly told you that is not the case. You ignore them and continue to claim things that aren't possible based on the statistics presented.

I suggest you start a new thread to 'discuss' scientific studies and interpretation. Maybe on a statistics forum or research forum.

 

[JayKosta]

Re: Re:

...
Again from Wikipedia article on statistical significance

The p-value is the probability of obtaining at least as extreme results given that the null hypothesis is true

...

------------------------------------
Yes, that is also my understanding.
And I think it also implies that there is a 1-p probability of obtaining some OTHER RESULTS if the null hypothesis is true.

Jay Kosta
Endwell NY USA

 

[JamesCun]

FrankDay:

However, that is how it is generally interpreted as these studies are designed to determine if a certain thing has an affect. If the differences are small it is reasonable to assume the groups are the same. But, you are right, it is not technically correct. Good to point out in your statistics class but perhaps a bit over the top for this forum. That is why we are relying on you to tell those of us who don't have your background how to properly interpret and talk about this kind of data.

KingBoonen:

It is not a small point, and it is not how they are interpreted. It is completely fundamental to any interpretation of a p-value and seems to be the basis of your misunderstanding. The populations can vary wildly, the null hypothesis is that any variation is due to chance alone.

Frank day:

That could only occur if the two groups were identical, yet above you said that wasn't required.

I think you are confusing things with multiple different terms. You are using groups to mean data sets, interventions and a bunch of other meanings. In a study that involves two different interventions, the results of the two interventions don't need to be identical for the null hypothesis to be true. If the statistics show that the results are likely due to chance, the null hypothesis is accepted. It doesn't mean the two interventions or groups are identical.

Edit: it also has nothing to do with the difference observed being small or large. The significance is independent of the delta between groups/interventions/data sets. What you've also missed from the discussion here is the other ways of viewing data from research studies. You can use subjective comparisons and look at averages, trends and raw data. That is all useful stuff when looking at what to do in future studies.

 

[FrankDay]

Re: Re:

...
Again from Wikipedia article on statistical significance

The p-value is the probability of obtaining at least as extreme results given that the null hypothesis is true

...

------------------------------------
Yes, that is also my understanding.
And I think it also implies that there is a 1-p probability of obtaining some OTHER RESULTS if the null hypothesis is true.

Jay Kosta
Endwell NY USA

It is not the probability of achieving SOME OTHER RESULT, it is the probability that the differences seen are not due to chance, in other words, something in the study, usually what is being studied if it is a good study design, resulted in the difference.

 

[FrankDay]

Re: Re:

And it seems that confusion about p-value is fairly widespread - see here for another discussion about its complexity.
http://blogs.plos.org/publichealth/2015/06/24/p-values/

Jay Kosta
Endwell NY USA

They are hugely misunderstood, misinterpreted and misrepresented, although this is true of many statistical values and methods (don't get me started on PCA). A lot of the time they're not even corrected! However it's extremely well known in science and there has been a steady push to improve on these matters as big data increases the requirements for statistics in areas of science where traditionally ratios and t-tests were the norm.

So someone on a cycling forum not really getting them is completely understandable, some of the smartest people I know don't really understand them. But they actually listen when people who do understand them tell them why they can't do what they are doing...

"They" don't listen to you because you don't tell them what they should be saying. I don't know what your academic background is but I suspect it is substantial. I would just like to point out that Dr. means teacher, at least that is what I was told in medical school. As I said, I see this way of "debating" as simply academic bullying. You point out where others are wrong (which of course makes you superior) but never say what is correct. I look forward to your eventually telling the group how one should properly interpret a p=0.125 between two groups showing a difference.

Frank, this has been explained to you over and over again. I'm quite surprised anyone is still willing to write anything with all of your bullying on this thread.

The interpretation is that the researchers didn't find a significant difference in their study. That is why they reported .125 and .25. If they had found a significant difference in the data, they would've reported that. What you are doing is misleading and constantly shifting your language to confuse the subject. You've repeatedly said that the p value is the probability of the null hypothesis being true. Others have repeatedly told you that is not the case. You ignore them and continue to claim things that aren't possible based on the statistics presented.

I suggest you start a new thread to 'discuss' scientific studies and interpretation. Maybe on a statistics forum or research forum.

I would suggest you actually read the question direct towards King. I understand that .125 does not reach the 0.05 (the arbitrary "significant" cut-off). The question is how to interpret that 0.125 number beyond that it didn't reach the cut-off? According to Wikipedia whatever the P-value is represents "the probability of obtaining at least as extreme results given that the null hypothesis is true" King has said the P-value has nothing to do with probability. I have asked him to come and clarify the situation since other usually acceptable sources seem to disagree with him.

 

[PhitBoy]

Re: Re:

Just to bring us back on topic: Lets not forget that the abstracted presented by the OP includes the result: "Metabolic cost for the four incremental stages was reduced by 1.2±0.1kcal/min (p<0.001)". Within the context of p values, that would seem to indicate that we can be pretty confident that pedaling with a counterweight is less metabolically costly than pedaling without one. This is true for cyclists who follow instruction to pull up more and for a National Paralympic Champion with 7 years of single leg immersion training.
Uncoupled cranks such as the original Smart cranks and other similar products train one to pedal in a way that closely approximates single leg non counterweighted cycling. Hence, uncoupled cranks train you to pedal with a more costly technique. Seems like something one would want to avoid.

 

[JamesCun]

Re: Re:

And it seems that confusion about p-value is fairly widespread - see here for another discussion about its complexity.
http://blogs.plos.org/publichealth/2015/06/24/p-values/

Jay Kosta
Endwell NY USA

They are hugely misunderstood, misinterpreted and misrepresented, although this is true of many statistical values and methods (don't get me started on PCA). A lot of the time they're not even corrected! However it's extremely well known in science and there has been a steady push to improve on these matters as big data increases the requirements for statistics in areas of science where traditionally ratios and t-tests were the norm.

So someone on a cycling forum not really getting them is completely understandable, some of the smartest people I know don't really understand them. But they actually listen when people who do understand them tell them why they can't do what they are doing...

"They" don't listen to you because you don't tell them what they should be saying. I don't know what your academic background is but I suspect it is substantial. I would just like to point out that Dr. means teacher, at least that is what I was told in medical school. As I said, I see this way of "debating" as simply academic bullying. You point out where others are wrong (which of course makes you superior) but never say what is correct. I look forward to your eventually telling the group how one should properly interpret a p=0.125 between two groups showing a difference.

Frank, this has been explained to you over and over again. I'm quite surprised anyone is still willing to write anything with all of your bullying on this thread.

The interpretation is that the researchers didn't find a significant difference in their study. That is why they reported .125 and .25. If they had found a significant difference in the data, they would've reported that. What you are doing is misleading and constantly shifting your language to confuse the subject. You've repeatedly said that the p value is the probability of the null hypothesis being true. Others have repeatedly told you that is not the case. You ignore them and continue to claim things that aren't possible based on the statistics presented.

I suggest you start a new thread to 'discuss' scientific studies and interpretation. Maybe on a statistics forum or research forum.

I would suggest you actually read the question direct towards King. I understand that .125 does not reach the 0.05 (the arbitrary "significant" cut-off). The question is how to interpret that 0.125 number beyond that it didn't reach the cut-off? According to Wikipedia whatever the P-value is represents "the probability of obtaining at least as extreme results given that the null hypothesis is true" King has said the P-value has nothing to do with probability. I have asked him to come and clarify the situation since other usually acceptable sources seem to disagree with him.

The interpretation of the study is that it didn't result in any findings that met the level of significance required by the authors. In other words, the authors didn't think the data was strong enough to conclude that the null hypothesis was false. Any other interpretation by people reading the study is interesting and would be ideal for developing the next study on the same topic.

 

[FrankDay]

Re: Re:

Just to bring us back on topic: Lets not forget that the abstracted presented by the OP includes the result: "Metabolic cost for the four incremental stages was reduced by 1.2±0.1kcal/min (p<0.001)". Within the context of p values, that would seem to indicate that we can be pretty confident that pedaling with a counterweight is less metabolically costly than pedaling without one. This is true for cyclists who follow instruction to pull up more and for a National Paralympic Champion with 7 years of single leg immersion training.
Uncoupled cranks such as the original Smart cranks and other similar products train one to pedal in a way that closely approximates single leg non counterweighted cycling. Hence, uncoupled cranks train you to pedal with a more costly technique. Seems like something one would want to avoid.

Well, if you had actually read this thread you would have figured out that neither of the two conditions tested have much to do with how ordinary people pedal whether they have trained with PowerCranks or not. It would be of great interest if one were an amputee cyclist but if one is not then ???

 

[FrankDay]

Re: Re:

The interpretation of the study is that it didn't result in any findings that met the level of significance required by the authors. In other words, the authors didn't think the data was strong enough to conclude that the null hypothesis was false. Any other interpretation by people reading the study is interesting and would be ideal for developing the next study on the same topic.

Thanks for that but that is not the question. Just what does P-value mean? Does it have any meaning if it doesn't meet the arbitrary "significance" level? Perhaps p value has nothing to do with probability but just happened to be the 16th value defined using an alphabetical sequence. Maybe King can tell us as he has held himself out as an expert in this area.

 

[PhitBoy]

Re: Re:

if you had actually read this thread you would have figured out that neither of the two conditions tested have much to do with how ordinary people pedal whether they have trained with PowerCranks or not.

Please remind me how single leg cycling without a counterweight is different than cycling with uncoupled cranks? I seem to recall something about how uncoupled cranks actually provide a counterweight but I couldn't really follow your logic. Could you help us all again with that?

 

[FrankDay]

Re: Re:

if you had actually read this thread you would have figured out that neither of the two conditions tested have much to do with how ordinary people pedal whether they have trained with PowerCranks or not.

Please remind me how single leg cycling without a counterweight is different than cycling with uncoupled cranks? I seem to recall something about how uncoupled cranks actually provide a counterweight but I couldn't really follow your logic. Could you help us all again with that?

Well, uncoupled cranks do provide a counterweight such that the rider doesn't have to pull up very hard to get the cranks up but the counterweight increases the effort the pushing leg must do for the same power since it not only has to push as hard but also push the counterweight up, a condition that doesn't exist in uncoupled cranks. Your problem is you don't have a clue what uncoupled cranks do or don't do because you don't have any personal experience with them. You sole experience with them is what you imagine them to be. Thought experiments can be useful but only when the thinking is correct.

 

[JayKosta]

Re: Re:

...
Again from Wikipedia article on statistical significance

The p-value is the probability of obtaining at least as extreme results given that the null hypothesis is true

...

------------------------------------
Yes, that is also my understanding.
And I think it also implies that there is a 1-p probability of obtaining some OTHER RESULTS if the null hypothesis is true.

Jay Kosta
Endwell NY USA

It is not the probability of achieving SOME OTHER RESULT, it is the probability that the differences seen are not due to chance, in other words, something in the study, usually what is being studied if it is a good study design, resulted in the difference.

-------------------------------
Edit - and this is ALL in the situation that the null hypothesis is TRUE
-
Don't over-think the 1-p probability --- consider rolling a single dice (having 6 faces with values 1 thru 6).
The probability of rolling a '1' is 1 in 6 (16.66%)
The probability of rolling SOME OTHER RESULT than a '1' is 5 in 6 ( 100 - 16.66%)

The definition of p-value is actually quite simple and understandable - the trouble comes when trying to twist it to have more meaning than it really does.

Jay Kosta
Endwell NY USA

 

[PhitBoy]

Re: Re:

uncoupled cranks do provide a counterweight such that the rider doesn't have to pull up very hard to get the cranks up

Yep this is where you lost me last time. Can you please sketch up a free body diagram that shows how the uncoupled crank has some sort of counterweight function?

 

[JamesCun]

Re: Re:

if you had actually read this thread you would have figured out that neither of the two conditions tested have much to do with how ordinary people pedal whether they have trained with PowerCranks or not.

Please remind me how single leg cycling without a counterweight is different than cycling with uncoupled cranks? I seem to recall something about how uncoupled cranks actually provide a counterweight but I couldn't really follow your logic. Could you help us all again with that?

Well, uncoupled cranks do provide a counterweight such that the rider doesn't have to pull up very hard to get the cranks up but the counterweight increases the effort the pushing leg must do for the same power since it not only has to push as hard but also push the counterweight up, a condition that doesn't exist in uncoupled cranks. Your problem is you don't have a clue what uncoupled cranks do or don't do because you don't have any personal experience with them. You sole experience with them is what you imagine them to be. Thought experiments can be useful but only when the thinking is correct.

Uncoupled cranks can never provide a counterweight. Please stop with these complete fabrications of reality.

 

[backdoor]

Re: Re:

if you had


Well, uncoupled cranks do provide a counterweight such that the rider doesn't have to pull up very hard to get the cranks up but the counterweight increases the effort the pushing leg must do for the same power since it not only has to push as hard but also push the counterweight up, a condition that doesn't exist in uncoupled cranks. Your problem is you don't have a clue what uncoupled cranks do or don't do because you don't have any personal experience with them. You sole experience with them is what you imagine them to be. Thought experiments can be useful but only when the thinking is correct.

Does this uncoupled crank "counterweight" remove the greatest difficulty in one legged pedalling that occurs as pedal is being drawn from 10 to 12 o'c, a one legged counterweight completely eliminates this waste of energy and that extra pushing down effort for the same power will improve a rider's performance when he returns to unweighting two legged pedalling.

 

[PhitBoy]

Re: Re:

Uncoupled cranks can never provide a counterweight.

That's what I thought too. Isn't that whole point of using uncoupled cranks? So each leg has to work independently to lift itself during the flexion phase? Like pedaling with one leg but you get to stay clipped in with both legs.
I'm interested to see Frank's free body diagram. Perhaps in trying to draw it he will see an error in his thinking.

 

[JamesCun]

Re: Re:

Uncoupled cranks can never provide a counterweight.

That's what I thought too. Isn't that whole point of using uncoupled cranks? So each leg has to work independently to lift itself during the flexion phase? Like pedaling with one leg but you get to stay clipped in with both legs.
I'm interested to see Frank's free body diagram. Perhaps in trying to draw it he will see an error in his thinking.

Absolutely. It goes against everything the product is intended to do. The legs are independent, any momentum created is no different than a tailwind or downhill, in terms of assisting with reducing power needed on the upstroke.

He has been asked repeatedly for this simple diagram. I'm assuming his refusal to do so indicates he is incorrect in his thinking and doesn't want to share that with the group. Too bad, since this is far more on topic than a discussion of statistical significance of a different study.

 

[JamesCun]

Re:

So, here is a new study trying to replicate Lutrell. http://www.researchgate.net/publica..._using_uncoupled_cranks?citationList=outgoing

Objectives: Uncoupled cycling cranks are designed to remove the ability of one leg to
assist the other during the cycling action. It has been suggested that training with this
type of crank can increase mechanical efficiency. However, whether these improvements
can confer performance enhancement in already well-trained cyclists has not
been reported. Method: Fourteen well-trained cyclists (13 males, 1 female; 32.4 ± 8.8
y; 74.5 ± 10.3 kg; Vo2max 60.6 ± 5.5 mL·kg−1·min−1; mean ± SD) participated in this
study. Participants were randomized to training on a stationary bicycle using either an
uncoupled (n = 7) or traditional crank (n = 7) system. Training involved 1-h sessions,
3 days per week for 6 weeks, and at a heart rate equivalent to 70% of peak power
output (PPO) substituted into the training schedule in place of other training. Vo2max,
lactate threshold, gross efficiency, and cycling performance were measured before
and following the training intervention. Pre- and posttesting was conducted using
traditional cranks. Results: No differences were observed between the groups for
changes in Vo2max, lactate threshold, gross efficiency, or average power maintained
during a 30-minute time trial. Conclusion: Our results indicate that 6 weeks (18 sessions)
of training using an uncoupled crank system does not result in changes in any
physiological or performance measures in well-trained cyclists.

Many of you will note that they show "no difference". Of course, there are differences but they just don't reach the P<.05 level of significance.
For instance: Gross efficiency in the PC group improved from 19.7 to 20.9 (a 6% improvement) while the control group improved from 19.8 to 20.3 (a 2.5% improvement). This difference only reached the 0.25 level of significance. So, there is a 1 in 4 chance this difference is due to chance or a 3 in 4 chance (75%) the differences are real.
Then, time-trial power. The PC group improved from 284 to 298 watts (5%) while the control group improved from 274 to 281 watts (2.5%). This difference only reached the 0.125 level of significance. So, there is a 1 in 8 chance this difference is due to chance or a 7 in 8 chance (87.5%) the differences are real.

Most of the uncoupled cranks studies that have "failed" to show a difference look about the same, there being a trend to showing a difference but not reaching the scientific standard of 0.05, or a 1 in 20 chance the result is due to chance so the author is forced by convention to say he found no difference.

As the authors of this study wrote in their discussion: "The lack of effect of training using uncoupled cranks on GE is in contrast to that of Luttrell and Potteiger. A potential reason for these disparate results may be related to the participant recruitment criteria."

So, here is the problem with interpreting these studies. Here we have a study in which the PC group shows an increased improvement in efficiency over the control group that has a 75% chance of being a real change. And, the PC group has an increase in power over the control group that has an 87% chance of being real. Yet, because the results don't reach the 95% chance of being real, that scientists have generally agreed is required to be shown before someone can claim their study to have shown a difference, Fergie and others feel free to state that PC's have been "proven" to be a failure and a fraud when, in fact, the data suggests the exact opposite. This scenario is pretty much the case with every study that has looked at the PC's, the trend is there but they don't reach the p<0.05 level so the author concludes, rightly for a scientific paper, the data shows no difference. The problem it seems is it is difficult to show this difference, especially in experienced cyclists, in only 6 weeks.

My guess is if someone were to do a meta-analysis of all the studies out there there is enough data now to demonstrate a difference in both power improvement and efficiency improvement to the scientific standard. Until then, the world will just have to accept that pretty much all of the studies have shown PC's to be effective, just not to the 95% confidence level. This seems more an issue with study design (not enough subjects, not lasting long enough, etc.) than with the concept itself. So, if one is satisfied with evidence to the 70-90% level then it is there. If one needs the 95% level, then you may have to wait a bit.

Frank, can you explain how this study you mention is related to this thread? It has now been almost 10 pages of off topic BS and faulty interpretations of statistics. I'm not sure why this is being allowed in this thread???

 

[Irondan]

Re: Re:

Frank, can you explain how this study you mention is related to this thread? It has now been almost 10 pages of off topic BS and faulty interpretations of statistics. I'm not sure why this is being allowed in this thread???

He can not. See member suspension thread...

 

[JayKosta]

Re: Re:

...
Uncoupled cranks such as the original Smart cranks and other similar products train one to pedal in a way that closely approximates single leg non counterweighted cycling.
...

-------
I thought that a single-legged cyclist without a counterweight would actually be trying to produce upstroke power that is around 30% (my estimation) of the overall average power. The desire (and need) to do that being to keep the speed of the bicycle fairly constant from downstroke to downstroke (to avoid a noticeable cyclic speedup - slowdown for each crank rotation) .
Single-leg with a counterweight allows the upstroke power that the cyclist applies to be reduced because of the momentum of the counterweight. And by reducing upstroke effort, the cyclist is able to extert more downstroke effort to produce the same average power AND STILL KEEP THE BIKE SPEED FAIRLY CONTSTANT.

Jay Kosta
Endwell NY USA

 

[Alex Simmons/RST]

Re: Re:

...
Uncoupled cranks such as the original Smart cranks and other similar products train one to pedal in a way that closely approximates single leg non counterweighted cycling.
...

-------
I thought that a single-legged cyclist without a counterweight would actually be trying to produce upstroke power that is around 30% (my estimation) of the overall average power. The desire (and need) to do that being to keep the speed of the bicycle fairly constant from downstroke to downstroke (to avoid a noticeable cyclic speedup - slowdown for each crank rotation) .
Single-leg with a counterweight allows the upstroke power that the cyclist applies to be reduced because of the momentum of the counterweight. And by reducing upstroke effort, the cyclist is able to extert more downstroke effort to produce the same average power AND STILL KEEP THE BIKE SPEED FAIRLY CONTSTANT.

Jay Kosta
Endwell NY USA

I think the amount of "upstroke" power would likely be a function of inertial load, and not nearly as high as you suggest. IOW at lower speeds then likely a greater proportion of power (but still minor) comes from upstroke, but lessens as speed increases.

Also, as speed increases, the variation in crank velocity and rider speed due to the pulsative nature of power application falls (the inertia of the bike+rider acts like a flywheel does in an engine to smooth out such velocity variations).

I demonstrated this in this post about the velocity variations of crank and bike during accelerations, albeit a two-legged model, but the principle is the same for any pulsative style of power application:

http://alex-cycle.blogspot.com.au/2015/01/accelerating-sins-crank-velocity.html

Note how little the velocity varies about its mean relative to the very large variations in power through a crank cycle.