Hacker News Reader

Comment by mlthoughts2018 | Hacker News Reader

mlthoughts2018

Yes, but I’m not sure what that would matter. I suppose they needed to control for time of day among marathons, but often they are all early on weekend days.

The point is to study the causal impact of having marathons vs not. Comparing that effect to other times of week which does not contain a marathon is not useful for much.

It could tell you if hosting a marathon is about as costly as having X more rush hours’ worth of mortality or something, but does not relate to the causal effect of the marathon.

For an example to compare, think about studying the impact on crime of heatwaves in a winter month. Someone could say, but won’t there be spikes of crime from heat in summer months? Yes, but why is it relevant — the point is to know the effect of the heatwave.

View on HN

Effect size and comparison to other causes matters, because every study will be used as evidence for some political argument.

This study ("Marathons kill people!") will be used to argue for banning marathons in cities by people why are inconvenienced by the road closures.

A comparison to other common and accepted causes of traffic delays - or positive health impact of marathons - will prevent this misuse.

A similar thing happens with wind turbines killing birds. Yes they do. But radio towers, power pylons, agriculture, pesticide use, cars each kill 100x more, building windows 1000x more, and cats 10000x more. Yet it is still a popular argument used by people who are against renewable energy for other reasons, while nobody argues for banning skyscrapers.

I agree in general and even said exactly this in my comment (comparing effect size in terms of X rush hours’ worth of mortality).

But in the specific case of policy adoption towards marathons, I don’t agree that comparison matters much for interpretation.

Banning marathons is totally different than taking action to reduce generic or rush hour traffic. Marathons are a purely recreational thing for society, nobody has any sincere need for them, certainly no needs to be colocated with dense urban areas.

Rush hour traffic is an emergent effect of necessary daily commuters and operations of a city.

Disentangling what part of it is needed, what can be reduced, and how to politically coordinate it is an entirely different situation.

In this sense, if banning marathons prevents even a tiny effect size on mortality, it’s obviously worth it, no matter if it is only saving 0.1 rush hours’ worth of mortality, or 0.01, or 0.001. Comparing with rush hour effect size adds virtually nothing to the process.

Banning marathons to save one life just because they’re recreational? Would you also ban all books because people have died driving to bookstores?

This is precisely proving my whole point.

The debate would be about the actual effect size as compared with the way society determines subjective value of things like marathons. It would have nothing at all to do with the relative effect size compared to some other benchmark like “rush hour.”

> If banning [something] prevents even a tiny effect size on [bad outcome] it’s obviously worth it

The same logic is why we now shuffle around in our socks at the airport. It is a terrible recipe for making policy.

Of course if you take it out of context it fails to generalize. Why would you think it should generalize?

Nobody is saying this about [something] or [bad outcome], only about marathons that demonstrate causal impact of higher mortality rate due to traffic impact on emergency services.

It’s not a recipe. It’s weird to me that you’d even suggest such a non sequitur concept.

Sure, presence of a non-zero effect, regardless of its size, can be theoretically interesting. Relativistic effects are often small, but they’re critical support for the theory of relativity. I’m not sure there’s an overarching theory here, but if you want to make that argument, be my guest.

On the other hand, the article has obvious policy implications, and those ought to depend, critically, on the size of the effect. If marathons are, in fact, massively risky, then maybe we should have them far, far away from hospitals. If the risk bops around a lot, and having a marathon shifts it from “quiet Sunday morning” to “run of the mill Tuesday” then no one needs to do anything.

> “If the risk bops around a lot, and having a marathon shifts it from “quiet Sunday morning” to “run of the mill Tuesday” then no one needs to do anything.”

I think you are assuming some kind of implied cost-benefit here that I don’t see, or at least why would it be delimited in units of effect size of other times of a week.

Even if the effect is as small as shifting quiet Sunday to run-of-the-mill Tuesday, it still seems obviously worth it to remove that. I’d say the same about parades and other things like this.

The debate would have no real connection to the effect size in units of other days’ traffic mortality spikes, ever.

The debate would be how do we value subjective recreation things like parades or marathons, to justify any effect size of mortality, and then how does that relate to the actual effect size.

Marathon vs. rush hour makes no difference to anyone. Why is rush hour the unit to care about? Marathon vs. no marathon assuming non-zero effect size of any kind is the discussion.

I brought up rush hour because it’s an intuitive benchmark.

If my relative risk changes a thousandfold over the day, then I don’t really care about a ~1 point increase; by living in a city, I’ve already tacitly accepted more risk. On the other hand, if it normally swings from 0.9-1.1, then the damn things are a menace (though still not a huge one in absolute terms) and maybe we ought to do something.

I still don’t agree. Why these thresholds?

What if the behavior in question wasn’t marathons but instead was attempts at the Guiness Book of World Records record for largest number of people twirling umbrellas in a line while blocking traffic?

Does that also need to show an effect size relative to rush hour or generic risk tolerance of some other situation? Why? Why would it be “worth” that particular amount of risk and not something much greater (if you love setting records) or much lower (if you think it’s totally pointless & blocking fire trucks for nothing).

Again, I brought up rush hour as a reference point (benchmark, calibration method, etc), not as a threshold. Surely you’ve seen the “Is four a lot?” meme (“Dollars? No. Murders? Yes”). That applies here in spades.

And yes, I think the bar for some random record attempt would be a bit higher. Some of the comments in this thread seem to think the marathon is valueless. I disagree. The Boston marathon is a big deal locally—-even non-runners get excited about it, so it does contribute to civic pride or whatever, More tangibly, the marquee marathons (Boston, NYC, etc) bring in a fair bit of tourist money.

If these things are unexpectedly very dangerous, then sure, let’s reconsider. If not, do you also want to replace Central Park with rounded, rubberized surfaces?

Reply on news.ycombinator.com