Statistical evidence that new shoe technology may not improve elite marathon performance by more than 30 s (whole distance)

you are not accounting for shoe type?

Double digit iq reasoning

How about doing a real study. Oh wait, they already have.

https://www.ncbi.nlm.nih.gov/pubmed/30374945

So, we need to know how typical a 30s change in these these times in a year is. I imagine that just due to randomness this average bounces around quite a bit year to year. We need to know how much to get a handle on whether 30sec looks like the signal or the noise. Another related question, is suppose there is a very fast or very slow race in on of the majors. How much would that be expected to skew this average?

Now calculated it separately for Nike and non-Nike athletes

Let me guess... you also don't "believe" in "climate change."

Bichard Ascroft wrote:

Double digit iq reasoning

Well if you believe the hype, then the new shoes are faster than running on a track in the lightest spikes.

And people do believe the hype don't they?

That’s the funny thing about “shoe technology”. The manufacturer has no interest in actually proving or disproving their claims.

If they can prove that it makes a big difference, the shoes might get banned from competition. If they prove it makes very little (or even NO) difference - they won’t sell any shoes!

A statistical sample of ONE. Really?

I posted a comment similar to this one on a different thread. The average top 10 time in 2019 was 2:03:16, which is 98.9% of the average average top 10 time from 2011-2017. I'm pretty sure all of these times where run in some model of the vaporfly. Does anyone have confirmation on that?

On the other hand, the vaporfly has not had nearly as big of an impact on any of the other seemingly relevant stats. Even though the vaporfly has been available since 2017 and I'm fairly certain all or most of the top 10 times in 2018 where set in them, the average top 10 time in 2018 was 2:03:58, which is only 99.4% of the average. Similarly, the vaporfly hasn't had much of an impact on top 100 times in general: the average for 2019 was 2:06:50 and the average for 2018 was 2:06:54, which are both 99.5% of the typical times for '11-'17. In fact, 2012 was deeper than either of these two years, with a very comparable top 100 average and a faster 51-100 average.

There's also the fact that if these shoes are worth 1% plus then we've seen a major setback in sub 2:04 and 2:05 performances. With this alteration only Kipchoge breaks 2:05 in 2018 and only Kipchoge and Bekele break 2:04 in 2019.

Of course in order for stats about the top 100 times to mean much we need to know what percent of athletes started wearing the vaporfly. Does anyone have this info? It seems like a pretty large percent, at least from the last few big races where non nike athletes got permission to run in the vaporfly and you need to go pretty far down the list to find the non vaporfly winner. Either way, the 2019 top 10 stat is the only big red flag indicating these shoes are worth more than .5%, and none of these stats seem to come close to suggesting a 2%+ effect like a bunch of posters seem to be claiming.

I did not deny that there might be a slight advantage, it just seems to be much less than many of you think.

Most of you based their opinion on the impression, that elite marathon times improved a lot. I was just curious and calculated how much. The result is delusional.

To the guy who wondered about statistical noise: there are quite consistent trends if improvement (2008-12), stagnation (2012-17) and very slight improvement (2017-19). There is a very low level of "noise". For that I used averages like top 30, not individuals like Kipchoge.

Of course some athletes still run in other shoes but not many. Of course other aspects like PEDs, training methods (and maybe the climate change like someone suggested) have to be taken into account.

Thank you for this.

I did use the top 30 average because the top 10 relies too much on individual performances and single fast races.

Something to think about: world top performances are strongly biased towards Kenyan/Ethiopian performances. Now changes of the situation in these countries like better / worse doping control in Kenya might have a high impact.

Maybe I calculate the trends for other areas like US or Europe later. If the trends are the same, it is more likely caused by a general change like new shoe technology.

Maybe there are even overlapping effects like a massive improvement due to shoes and decreasing performances due to less doping in Kenya. Evaluation of other countries / regions might help to filter effects like this.

I still believe the pacers blocking wind the whole way is a greater advantage than the shoes. If it wasn't, they would have Kipchoge run the marathon on his own. Of course we know he would be lucky to break 2:04.

Over 13 miles per hour...... wrote:

I still believe the pacers blocking wind the whole way is a greater advantage than the shoes. If it wasn't, they would have Kipchoge run the marathon on his own. Of course we know he would be lucky to break 2:04.

greater? Maybe, maybe not. But I don't think there is any serious doubt that illegal pacing offers a significant advantage, likely worth approximately 2 minutes.

I made this similar analysis looking at the top marathon times (like total performances under 2:06, and under 2:10) dating back to the late 2000's up through the past year, and looking at the numbers before and after the release of the vapor fly 4%.

I would give at the very most a 60 second advantage to the era after the shoe was released in roughly 2017 (some people got them slightly before the rest of the public). But really I would say it's negligible and far lower than 60 second advantage over a full marathon. There was no obvious improvement in the performances as a whole, and would need a larger sample size. And a larger sample size could come in more time, but there are too many other factors that we can't keep consistent (like larger population, and more people running more) to make a real conclusion.

0) Did you include outliers, such as Boston 2018? Did you check to see if including affects your results?

1) Did you calculate dispersion/standard deviation? Yearly times bounce around a lot, so we would expect any year's times to be different from the ten year average. Unless we know how times vary, we can't tell if the difference between any year and the average is noteworthy.

2) Did you account for trends? If times are getting lower over time, ten year averages are pretty much meaningless. We need to know how this year's times differ from those we'd expect to see. And what we expect to see depends on the trends.

I can't wait to see what you come up with.

Funny you Americans are so focused on Boston. It is not record eligible and thus not included in any top list. Even if it was a regular course it is most times won in 2:09 or thereabouts, so of no significance for calculations based on top 30 or even top 100 performances.

The top performances like top 5 or 10 are scattering because these are too dependent on a small number of individuals. However, averages of top 30 or higher numbers showed a high consistency and very low scattering.

The year 2015 was a clear negative outlier, like more than a minute slower than the years before and after. Any idea how this might have happened?

Run a model with time as the dependent and percentile for that year as the independent variable. Also include year as well as an indicator variable to control for the years post vapor fly release. Report back the coefficient estimates standard errors and measures of fit.