No statistically significant improvement in marathon results since Vaporfly introduction

nv4 wrote:

Sesamoiditis wrote:

You are not even using the term "sample size" correctly. Your sample size is 12 years. Based on your own reasoning that is too small.

No, sample size is correct, it's all 2921 results in history to date. I just posted statistics for the last decade or so but the distribution is correctly calculated. Here is full list of the same sample. It's actually interesting that 2008 have been a significant turning point, much more impactful than VF introduction with sudden significant increase in sub-2:10 performances.

Your are looking at number of sub 2:10 marathons per year to see if there has been a change with the introduction of vaporflys. Your sample size is number of years. If you were looking at a the number of sub 2:10 marathoners wearing vaporflys your sample size would be 2921 but you are not doing that. While the info you have posted is interesting you are hurting your argument by using scientific terms like sample size and statistically significant incorrectly in addition to not taking into account any other factors that could contribute to sub 2:10 marathons like weather.

Bad Wigins wrote:

sbeefyk2 wrote:

Some tracks are extra bouncy relative to other tracks and make you feel and run faster.

Since when does a bouncy track make you faster? Does it bounce you forward, or up? Where's the advantage in prolonging footstrike and lowering cadence?

Does bouncing save you energy? If the ground beneath you accelerates upward, exerting a greater force on your foot, do you not have to counter that with an equal and opposite greater force?

If the track is elastic, then it will be deformed by your impact in the downward direction and in the shear direction backward based on the force of your foot on the ground. Then, if there is elastic return that accelerates your foot during the time that your foot is in contact with the ground, the direction will be the resultant of the upward and forward direction. You're not going to find a rubber or foam material that can be deformed in one direction but not an orthogonal direction as well. So, the same philosophy applies to the shoe.

My bad, I am not a native English speaker so usage of terms might be off. My goal was to provide insight based on statistics. I think, provided how limited data is available, it's interesting comparison. It's funny though that I am being bashed for trying to provide statistically based conclusion yet there is little critic addressed to those that speculate on basis of a single case like Kipchoge and Bekele. I think it would be fair if each side would be asked to provide arguments on which they base their opinion

nv4 wrote:

Why is method not solid? Because you don't see what you would like to see?

Let's take three other successive years before VF era began:

2013,14,15 = 536 (18.3%)

2012,13,14 = 566 (19.3%)

2011,12,13 = 544 (18.6%)

2010,11,12 = 529 (18.1%)

These years are the last corrupt Diack years, pre Seppelt, with at best rudimentary testing in Kenya and Ethiopia who dominate these numbers. All of that has changed 2016 - 2019.

You are maybe just showing that unlimited doping vs. more restricted doping has a similar impact on sub-2:10 performances like VF vs. non-VF. But there may be other factors as well.

nv4 wrote:

It's funny though that I am being bashed for trying to provide statistically based conclusion yet there is little critic addressed to those that speculate on basis of a single case like Kipchoge and Bekele. I think it would be fair if each side would be asked to provide arguments on which they base their opinion

Indeed Kipchoge's 2:01:39 from 2018 is not that special, considering it improved the 2014 WR by 78 seconds. That's 19.5 seconds per year (19.5 s/a).

Let's compare that with other WRs of the last 20 years:

(da Costa 1998 2:06:05)

Khannouchi 1999 2:05:42 (23 s, 23 s/a)

Khannouchi 2002 2:05:38 (4 s, 1 s/a)

Tergat 2003 2:04:55 (43 s, 43 s/a)

Gebrselassie 2007 2:04:26 (29 s, 7.3 s/a)

Gebrselassie 2008 2:03:59 (27 s, 27 s/a)

Makau 2011 2:03:38 (21 s, 7 s/a)

Kipsang 2013 2:03:23 (15 s, 7.5 s/a)

Kimetto 2014 2:02:57 (26 s, 26 s/a)

Kipchoge 2018 2:01:39 (78 s, 19.5 s/a)

You notice that of the nine WRs in the last 20 years, three had a larger improvement per year than Kipchoge.

Or look at decade by decade: last decade we saw an improvement of 1:43, this decade 2:20. Better, yes, but not exactly groundbreaking.

This decade looks even less impressive on the ladies' side (mixed record):

From a 2:20:43 in 1999 to a 2:15:25 in 2003 to a 2:14:04 in 2019,

so from 5:18 in 2003 (318 s, 79.5 s/a) to 1:21 in 2019 (81 s, 5.1 s/a).

And yet newbies reacted shocked to the 2:14:04 - imagine our disbelief in 2003 (when shoes weren't even mentioned). I'd say it was overdue.

That's a good and interesting take. If you will have more ideas on how to look at possible (non)impact of valporflys, share it again. Don't pay any attention to the trolls.

yyerqt wrote:

It's because many athletes retired at the time of the Vaporfly introduction, because they felt they gave an unfair advantage. Had those people been less ethical, the amount of sub 2:10s would have increased by a lot.

Can you even imagine the simpleton who wrote a comment like this.

This thread is the first decent statistical data I’ve seen on the Vaporfly and only on Letsrun could data be refuted based on ‘evidence’ like that.

This is wrong. The approach taken is scientifically sound.

Factors like weather and different courses are controlled for in the fact that they are not discriminated. The same courses are run every year and the weather differs between races to an extent that the average across a year makes no statistical difference considering that all marathons are included.

It is extremely unlikely that one year the weather at every marathon raced was much better than all the races the next year.

The sample population is appropriate for the hypothesis also; there is no point looking at 2:30 marathon runners as most are unlikely to buy the shoe.

What would improve it is to increase the sample size of elite athletes by including women also.

Also, you could better control for other factors by comparing against the sub or non-elite and see if the gap has grown between the number of elite finishers under 2:10 and the number of finishers under 3 hours over time. That would control for population increases and weather/courses more accurately.

A step change in 2017 should be visible if the shoes have a statistically significant advantage.

If someone can get me a list of every

The number of sub 2:30 times for women by year (including all performances per year, not just top per athlete):

2019: 256

2018: 236

2017: 212

2016: 176

2015: 194

2014: 154

2013: 151

2012: 201

2011: 160

2010: 129

2009: 123

2008: 123

Liverpool FC wrote:

If someone can get me a list of every

...sub 2:10 instance and the shoe the runner was wearing then I can carry out the analysis.

Another way, is to take the first 5000 runners times from Chicago and then the 5000 from the last 3 years, as long as the weather is similar, and then find the average time. You won’t be able to control for the shoes though...

The average finish time in Chicago in 2015 was 4:18 for M40-44 (most likely age group to but VF or Next%)...anyone find the average for 2019 for the same age group?

exrunner wrote:

This thread is the first decent statistical data I’ve seen on the Vaporfly

Except it's not. This is statistical data on marathon times with a spurious claim that Vaporflys are the reason for recent trends in improvement (which aren't even particular statistical outsiders).

nv4 wrote:

this would give significant increase

You keep saying significant, but have done no analysis that shows significance (although it appears there would be no significant difference).

If 7 from 10 of the fastest marathon times are by people wearing the VF or Next% then a proper analysis (Spearman’s) would certainly show a positive correlation.

Liverpool FC wrote:

If 7 from 10 of the fastest marathon times are by people wearing the VF or Next% then a proper analysis (Spearman’s) would certainly show a positive correlation.

That's all it would show. There's no data included to confirm if the shoe was distributed en masse to hundreds of runners since inception or even if they were wearing it in the race.

The shoe was released in Spring of 2017. The Handmaid's Tale was released at the same time so maybe a bunch of pro runners in Kenya are sunning faster since throwing off the yolk of the patriarchy (lol).

The numbers are interesting and I thank the OP for putting it together, but drawing any conclusion from such a simple sample is basically textbook What Not To Do.

Racket wrote:

Liverpool FC wrote:

If 7 from 10 of the fastest marathon times are by people wearing the VF or Next% then a proper analysis (Spearman’s) would certainly show a positive correlation.

That's all it would show. There's no data included to confirm if the shoe was distributed en masse to hundreds of runners since inception or even if they were wearing it in the race.

The shoe was released in Spring of 2017. The Handmaid's Tale was released at the same time so maybe a bunch of pro runners in Kenya are sunning faster since throwing off the yolk of the patriarchy (lol).

The numbers are interesting and I thank the OP for putting it together, but drawing any conclusion from such a simple sample is basically textbook What Not To Do.

You’re correct of course. Until someone wants to study it properly with control groups and placebos and identical ? twins, then we’ll only technically be guessing.

I have witnessed several stalled runners smashing PB’s, more in the Next% than the VF though. Might be due to global warming though, but I’ll stick with the notion that the foam helps.

Racket wrote:

Liverpool FC wrote:

If 7 from 10 of the fastest marathon times are by people wearing the VF or Next% then a proper analysis (Spearman’s) would certainly show a positive correlation.

That's all it would show. There's no data included to confirm if the shoe was distributed en masse to hundreds of runners since inception or even if they were wearing it in the race.

The shoe was released in Spring of 2017. The Handmaid's Tale was released at the same time so maybe a bunch of pro runners in Kenya are sunning faster since throwing off the yolk of the patriarchy (lol).

The numbers are interesting and I thank the OP for putting it together, but drawing any conclusion from such a simple sample is basically textbook What Not To Do.

OK so I just realized I read the title of the thread wrong so I'm an idiot for that. However, my point still sort of works in the opposite direction because it doesn't prove Vaporfly's don't improve individual marathon performance, just that general times across the board haven't improved outside of the mean.

Liverpool FC wrote:

If 7 from 10 of the fastest marathon times are by people wearing the VF or Next% then a proper analysis (Spearman’s) would certainly show a positive correlation.

Not necessarily. They might wear them because of their sponsor, and said sponsor might sponsor the best runners because he is the biggest sponsor by far.

And the marathon gets faster all the time, with no apparent peak coinciding with these % shoes.