American Runners Have Never Been Slower | Hacker News Reader

American Runners Have Never Been Slower

runrepeat.com

71 pointsnotdang59 comments

'It has been a popular belief that the increase of the average marathon finish time is due to the fact that participating in running competitions is gaining popularity. With the popularity increase, the number of not so fit participants rises, and the average finish time increases.
This argument is not exhaustive.

Let alone that in the last two years (2015 and 2016) numbers of participants are declining and the finish time in all the four major race distances is still growing.'

This is a wrong argument. It is enough if just running in general gains popularity. If you take races as a sample from the running population the sample size does not mean that much - the average/median/percentile performance of the sample (if it is representative) may follow the performance of the whole running population!

This is the case for example if the racing population is a uniform random sample. But it may even happen that nowadays larger proportion of slower runners are going to races - because races are much more advertised/known for the hobby joggers than in the past, while elite runners were there always...

I feel like this counter-argument is glossed over and not thoroughly refuted. The stats go back to the 80s and yet it's not till the last two years that participation starts to decline? I think it's very likely that more and more amateurs are signing up for smaller races that didn't previously exist. There's probably enough variance that one year of decline is not enough to prove that participation increase is not a factor.

Also, I think there's a chance that the running races are losing some of their top athletes to triathlons which are also growing in popularity.

Triathlons, off-road running races, obstacle races (tough mudder, etc), and other races that aren't plain running.

Local to me, there are also a few extreme distance races (longer than marathon), usually on hiking trails.

Indeed - it is not implausible to suppose that the set of people who think 'maybe I could run a marathon' has expanded over the years, and that expansion would inevitably be predominantly in the direction of including slower runners (BTW, not all slow runners are unfit.) If so, the average performance could decrease over time for this reason, even if the size of the subset that actually do choose to run a marathon has been falling lately.

Interesting claim but I'm not quite convinced based on the data presented. They considered the hypothesis ('myth') that increased popularity of running is what's slowing down the average times by looking at 100th-finisher times and nth-percentile times, but this does not control for an increased number of _races_ being held each year to accommodate all the extra runners.

It doesn't need to control for that. The new people who get swept up into participation in running events are invariably the slow ones. Conducting larger and more frequent races to fit them in is neither here nor there.

In a road race held somewhere in America in 1975, it would all have been largely highly fit, fast people from a track and field background.

It definitely needs to be accounted for. In the past decade, there has been a proliferation of smaller marathons. If you made a plot of "100th finisher times" by # of participants, it would be slower for smaller races. If the growth in marathons is primarily on the smaller end, then you'd expect the average to go down.

The "Playing Devil's Advocate" section takes this into account and simply orders all finishes for the year. This still shows the time increase.

> We analyzed the finish times of the 100th, 1,000th and 10,000th slowest participant (separately for men and women) for every year.

The article does not take into account race size - it plotted 100th finisher times by year, which is not the same as plotting 100th finisher times by # of participants.

If more races are held next year to accommodate more total runners, and the new runners are slower, and the number of races a given runner races each year stays the same, the average time of the nth finisher across any individual race would get slower.

Interesting, the goal of the fastest runners must not be to finish highly, if it were, they would distribute themselves to the additional races so that they had a better chance of winning. Instead, the claim is that they each keep running in the same races as they did before so that the slower runners do not have lower finishing places. Wait, not that either since then those new low finishes at the new races would be slower.

I note that they don't look at the top 3 finishers of big races... because those times have been falling (like the US Nationals). https://en.wikipedia.org/wiki/USA_Marathon_Championships

An economist might say that was impossible (for competitive people to deny themselves a precious placing goal), I simply find that it's unlikely.

The top runners will continue to go to the major races (New York, Chicago, Boston) because they can compete with the best runners in the world, there are appearance fees for elite runners, and they offer better exposure for athletes who depend on sponsorship dollars to make a living.

As for the US Marathon Championships, America's best marathoners usually skip that race (aside from when it serves as the Olympic Trials) because smaller, less prestigious marathons that pay less prize money are often chosen to serve as the US Championship. Why (relatively) smaller marathons? So that your US champion actually wins the race (like they would at the Twin Cities Marathon) instead of finishing somewhere in the top 5 (but not first) at New York.

So I can't speak for every runner, but for the vast majority placing will not be a major factor. I ran a half marathon recently and came in something like 600th - but that number is largely academic I was only really competing against the clock.

That race had several thousand entrance, of which maybe 3 had a realistic chance of winning and another 10 of placing. The things that attract people to races are more likely to be the location (either somewhere convenient or somewhere nice to run), timing (how does it fit in with training) and prestige of the race.

Interesting the prestige of the race means your position is likely to be much worse. But distance races, people don't ask where you came, they ask how quickly you ran.

That's one big thing I was worried about. Also, "We included only races with an average number of finishers greater than 2,000 (for all four distances)" could be a major confounding factor. What if many "competitive" runners switched to smaller events as big events became more "general"? Until this major issue is addressed, the conclusions can only apply to the self-selected set of American runners in large races.

Can't popularity explain this? Marathons used to be for an elite of first movers, then everyone joined and the elite left to do ironman and other sorts of crazy extreme.

I mean, this is a personal theory that I've basically pulled out my ass right this instant, based solely on what I've seen my coworkers do, so it's probably completely ridiculous. But in my network the people who were the fastest at running marathons aren't running marathons anymore.

Yeah I think this is also a possibility

(to be honest, healthwise, Marathons are very bad exercises - they're extremely biased towards aerobic capabilities - just check the amount of muscle mass a marathon runner has)

Also running for 3h straight is extremely boring and repetitive

This argument doesn't make much sense to me. Running is hard on your body sure. Lots of load on muscles and joints.

But then you use as argument 'extremely biased towards aerobic capabilities'. This argument doesn't make any sense. People have only very limited capability of anaerobic exercise. At the same time people are build to exercise aerobically almost forever.

> People have only very limited capability of anaerobic exercise. At the same time people are build to exercise aerobically almost forever.

Hence the need for balance

(though sports at the Olympic level are rarely healthy, I'm betting on the swimmers, short range runners (not 100m though), rather than the marathoners to be the healthiest)

Running a marathon is risky because of the load on muscles and joints. That's specific to running. Doesn't have anything to do with extreme aerobic exercise.

If you replace running with equivalent amount of cycling then there are no ill effects. You can easily cycle at a recreational level for 8 hours a day and see no ill effects. Muscle structure will be quite similar to distance runners. They just don't wear out as much.

> If you replace running with equivalent amount of cycling then there are no ill effects. You can easily cycle at a recreational level for 8 hours a day and see no ill effects. Muscle structure will be quite similar to distance runners. They just don't wear out as much.

I once thought this too, but I have come to think that this is a myth. It's what my doctors and physical therapists (and friends) say -- if you're injured running, pick up cycling.

I injured my knee cycling and haven't been able to do it in years. It's a chronic pain that starts when I hop on my bike, and only then. (I have had the full fit, bike sizing, and all that.)

I began running instead and have never felt better. I do it barefoot, but I don't know if that's relevant.

It's now hard for me to accept that cycling is easy on one's joints. Maybe it is true, but if it is true, it's not obvious. Our bodies are very good at running and walking. There is no reason to think they are good at cycling. A bike is a rather unusual and modern invention.

I can speculate that maybe it's the shoes that are contributing to injuries in distance runners. I am not sure. Feel free to reply if you feel I am wildly off.

Depends on the person. When fit I can enjoy running. I cannot explain why but (again, when fit, heart, muscles, joints all on appropriate state) it feels like flying.

3 hours straight? If only. In my previous marathon running past, if I made it in 4 hours I was ecstatic.

Must say that I disagree with you about it being boring - the marathons I took part in were more often than not set in beautiful locations and that was most of the enjoyment for me, seeing incredible landscapes.

Would you say that going hiking for 3 hours is also boring and repetitive?

Repetitive, sure. Boring, well that depends. I haven't run 3+ hours straight in going on 30 years, but I didn't find it so.

As soon as I saw the headline, I thought to myself "well yeah, Americans have also never been fatter," so the conclusion wasn't much of a surprise. Interesting that it's affecting men so much more.

This isn't the only study that indicates fatness is impacting aggregate performance, either. The US DoD has actually identified American obesity as a potential threat to national security and readiness. [1][2]

[1] https://www.usnews.com/opinion/blogs/policy-dose/articles/20... [2] (PDF) http://www.dtic.mil/dtic/tr/fulltext/u2/a547350.pdf

> I thought to myself "well yeah, Americans have also never been fatter,"

Do fat Americans run?

Some do.

But "fatter" != "fat". http://www.webmd.com/men/news/20041027/average-weight-for-am... points out that Americans are over 10kg heavier than the previous generation, but that's classified as "overweight" not "obese".

It's not a discrete scale, ceteris paribus every single kg will make you slower and increase injury risk. It's continuous.

If you wish to phrase things that way then feel free to reinterpret the discussion chain thusly:

1) spyhi used a continuous scale: "also never been fatter"

2) _qhtn switched to a discrete scale: "Do fat Americans run"

3) I replied to that question, that yes, "fat Americans run", objected to _qhtn's discretization of a continuous scale, and gave a pointer to a paper regarding BMI, which assigns categories like "overweight" and "obese" to the continuous scale.

Your comment comes across like you think I didn't know the difference between continuous and discrete, when that was the point of my response.

Picking up arbitrarily trends - obesity in this case, even though plausible, still doesn't feel right, ie. you could also choose average global temperature or %-age of CO2 and it would probably match the trend as well.

Also they're not addressing more complex implications - ie. if %-age of women (who are slower) increases and directly it doesn't account for total slowdown (as they've shown) maybe it has indirect impact on other runners (males) who are running _together_ with more slower participants (women).

I think assumption that people tend to run at the paste of people around them is not completely out of place compared to assumption that obesity slows down marathons.

(This comment references bodyweight. While not normative, some readers may wish to skip it.)

I am usually quite critical of misreported findings but I must say the article does seem to support its findings pretty clearly. If they had a plot of it, I would expect to show over this period another effect: "amateur marathon finishers have never weighed more, or been slower." Obviously this is my speculation and if found as I predict it would be a statement that even those who care enough about their health to train for, run, and finish in a marathon, have been weighing more and more.

I say this as a prediction because it is a very strong effect throughout the population. It certainly trumps any other health changes that have affected the population as a whole over the same period.

This data may be possible to estimate retroactively, visually based on photographs and footage of races - telling 15% from 25% from 35% from 40% bodyfat (for both genders) may be possible based on photographs alone. (With low fidelity.)

For comparison, although I couldn't find a chart going to 2017, this article includes weights for Americans as a whole through 2010:

https://www.washingtonpost.com/news/wonk/wp/2015/06/12/look-...

It is interesting to compare with the data we have just seen.

EDIT: I have edited this following a downvote. If there are specific objections I can edit it further or remove it.

I thought this study was going to be about the top percentile. I would assume this percentile is getting faster (competition is increasing in all sports at the top level).

Obviously on average Americans are getting slower because they are getting fatter. This almost isn't news.

Through, average Americans don't run marathons at all. It is only self selected smaller subset that does it (or is even able to finish it).

I don't understand why they rule out the increasing proportion of female runners. The fact that the trends are different for the genders doesn't negate this as a possibility. And they even state that this accounts for 46% of the effect. To me this reads more as an agenda-driven investigation than an actual study.

Both men and women are getting slower. Men much faster than women. So: Americans are becoming slower.

The myth tries to be the explanation for the average running time getting slower. It falsily implies that because more women join, people aren't really getting slower. This is false: people are getting slower.

I don't have the data, so the authors might be right, but honestly reading this that's the best gloss I can put on it.

The second last chart (they aren't numbered, because this is a study done by someone with a mathematical analysis Ph.D and a statistician, apparently) says it all - we've plotted two things and we have fitted a line and then we state a conclusion. I'm sorry, but stop the seminar there.

The first two points, what does the line look like if we remove those? What do you mean by average? Can we have error bars on the points to cover 90% of the population? What are the distributions of the finish times? If they are normal, how come - aren't there elite and club peaks as well as everyone else? Did we try excluding the registered runners?

The questions are endless and this study is "not exhaustive" - back to R guys.

Hey look, I'm part of a trend! :D

But seriously, I'm a bit confused. Early on, they refute "Just the slow are getting slower" but near the end they make almost the same claim. The trick they play is to shift from examining the N fastest to examining the N slowest. In other words, the two samples "prove" nearly contrary things because they're both incomplete. Also, the real meaning of N fastest/slowest changes a lot with race size. Using percentiles would have been a better (and more obvious) choice.

BTW, I love RunRepeat and have used it extensively as part of my research the last several times I've bought running shoes. Highly recommend.

Isn't this a good example of why people shouldn't just crunch through a big dataset looking for a correlation to support a claim that they already believe is true?

My first thought was also that it's simply due to the growing popularity of running. However this is also the case in Europe.

I've checked some random results - the difference is crazy. In Europe the median time seems to be around 1h55, while in the US ones it was more like 2h15-2h20 ! Which is very bad indeed

What times are those? They are not marathon times.

Male marathon world record is 2:02:57 and female is 2:17:01. It is unlikely that median marathon time in Europe would be faster then world record.

Marathons were cool in the 70s and 80s. Now athletic people are into different sports.

All the pop science I've read in the last decade has been saying the same thing: running on roads will screw up my knees, that long distance running is not great for either losing weight or gaining muscle, and that you should do intense running instead.

Slightly off topic, but this is a great example of an affiliate site done right. It's comprehensive, unbiased and has some really interesting content.

brainstorming: Has timing got more precise over time? Is it something in the atmosphere that's changed over time? Has the threshold of pain willing to be endured been reduced in our culture? Perhaps there is better awareness about the downsides of pushing through pain and injury.

Given that they haven't done random sampling of the US population over time, the correct title should be "population of american runners who choose to run marathons are getting slower."

The results could simply reflect the increasing popularity of participating in running events. The influx from new people is going to drags the performance averages down.

Quoting from the article: "It has been a popular belief that the increase of the average marathon finish time is due to the fact that participating in running competitions is gaining popularity. With the popularity increase, the number of not so fit participants rises, and the average finish time increases.

This argument is not exhaustive.

Let alone that in the last two years (2015 and 2016) numbers of participants are declining and the finish time in all the four major race distances is still growing."

https://news.ycombinator.com/item?id=14862901

Yes, I agree there are reasonable objections to the argument made in the paper. I think it's better to raise those objections than repeat a statement which the article explicitly addresses and dismisses as a myth.

You have to look at where the attrition is coming from; what is the demographic shift in the running population. Also, everyone is a year older with every passing year.

If you want a sport to stay vibrant, you have to have the fit youngsters coming in at the bottom, and the old people retiring out the top. This is not controlled in any way in public mass participation sport like in pro sport.

Really, this whole thing is meaningless. When I first saw the headline, I thought it would be about competitive running at the collegiate level or whatever.

Nope! Clickbait about changing jogger demographics.

What you say now is true, and not analyzed in the article.

It is not the same as your original statement regarding a simple increase in popularity.

Why does everything have to be a race issue?

I can't say that the study fully convinces me (I agree with szemet, not all counterarguments are carefully refuted).

There is however, a very different standard in the US, compared to other countries. In the Netherlands (and other countries), I am considered to have a normal or even slightly athletic posture. In the US, people consider me to be very skinny.

Aren't marathon runners mostly competing against others in the same race, so if they can all slow down, they all will? Some of the olympic races seemed like a farce with runners right next to each other - obviously adjusting their pace according to each other, not trying to set a record.

That's certainly true, and pacing can play a huge role in longer races. But, that also assumes that top runners are willing to out-sprint their competitors somewhere approaching the finish-line (not necessarily a full-on sprint, but even the last 5k of a marathon). We see this regularly in cycling, where aerodynamics and drafting play a huge role (and races are frequently 5+ hours). I'm not sure the same holds true for running, where even the longest race (marathon) is 3 hours and the draft is much less of a concern.

I think that applies to the top runners, who are competing with a few other specific individuals, but a typical marathon runner will be competing against the clock, if anything. In a race in which no participant is paying any attention to any of the other competitors, they will each still end up running with others going at about the same speed, assuming their pace does not fluctuate erratically (which it doesn't - especially in longer races, where almost all participants have practiced enough that they have an intuitive knowledge of what they can realistically expect to achieve.)

Nice thing is that they claim p-value of 0.01, much better than usual 0.05.

Healthy people don't run marathons

Just curious, but what cherry picked study and unique definition of health are you basing that statement on?

Reply on news.ycombinator.com