I don’t think climbing or swimming are as dangerous as driving.
They’re a couple orders of magnitude riskier, actually. It’s tricky to make a direct comparison because the risk of driving is usually expressed over distance traveled, while sports is usually measured over number of sessions, but if we assume a typical day’s driving is about 50 miles (80 km), then we’re looking at 0.1 micromorts per session, as opposed to 17 for swimming or 3.1 for rock climbing.
(I’m not totally sure I trust that swimming estimate. The one for rock climbing aligns with my intuition, although there’s a lot of variance within the sport—bouldering is comparatively safe, while attempting the world’s highest peaks is absurdly risky by sports standards. I did know one guy who died in a shallow-water blackout and none who died climbing, for whatever that’s worth.)
[ETA: The estimate for swimming turns out to be bogus. See below.]
The link you gave puts car deaths above swimming in the second diagram. It doesn’t say that the sporting numbers are measured by session. (Except for the BASE jumping, hang-gliding, scuba diving, canoeing, or rock climbing). My own research (the first three links from Googling “risk of car accident death”) puts car accidents consistently higher than swimming deaths.
I believe that’s because people drive much more than they swim, and the risk communication scale uses, say, your second numbers, and the comparison the link author gave converted that from annual to per-act.
I was trying to show that the swimming estimate wasn’t per session. 1 in 56,587 is close enough to 1 in 83,534 that they’re probably measuring the same thing, namely yearly deaths, in which case (assuming most swimmers swim more than 20 times a year, which I think is reasonable), the per-session risk for driving is more than that for swimming.
You’re right, it’s not per session—but it isn’t per year either. On closer examination it looks like they’re calculating the risk of death over the ten years surveyed (unless the 31 deaths reported are annualized, which I don’t think they are), which is an absolutely terrible bottom line—but fine, it makes the annual risk of death 1 in 566,000. I also notice that the population estimate is identical to that for running and cycling, so it’s probably some sort of very crude estimate of Germans involved in sports. Ugh. At least the climbing stats look more reliable.
Incidentally, an annual risk of death of 1 in 566,000 and a hundred sessions per year (two a week with time off for good behavior) gives us a per-act risk of 0.017 micromorts, about equal to driving four miles in a car.
It’s definitely not the chance of death in a year of swimming. My link already gives us all the numbers we need to calculate that—the number of deaths overall, the number of years being examined, and an estimate of the population involved—and it comes out to a chance of 1 in 5,658. (1,754,182 people / (31 deaths / 10 years).)
This conveniently lets us infer how they’re probably calculating the risk—it looks like they’re assuming one hundred sessions per year (or about two a week; fair enough) and doing a per-session estimate based on that. I also notice that the population estimate is identical to that for cycling and running, so it’s probably some sort of estimate of the number of people in Germany involved in an arbitrary popular sport. Cruder than I’d like, but I was only shooting for an estimate good to within an order of magnitude.
assuming most swimmers swim more than 20 times a year, which I think is reasonable
Those numbers look like general population numbers (and since it looks like a lot of drowning deaths are due to ineptitude, it seems unclear to me whether the yearly risk for frequent swimmers is higher or lower than for non-frequent swimmers). Instead of ‘all drowning,’ the 1 in 83,534 number, one should probably use the ‘in swimming pool’ number, which is 1 in 452,738.
I’m not sure I trust these estimates—or, rather, I don’t think I find them useful. The main problem is that the probabilities involved are all strongly conditional.
Consider swimming in a hotel swimming pool with a lifeguard watching and long-distance swimming alone in the ocean. Both are “swimming” but these two activities are radically different from the risk perspective. Similarly, you can do “climbing” in the climbing gym and you can do “climbing” in the Himalayas.
Sure, there’s a lot of variance involved. But there are more and less safe driving habits, too, and I’ll bet the variance is about as high. The point isn’t to demonstrate that one practice is under all conditions more or less safe than another, it’s to compare their average dangers as they’re actually practiced. And that clearly favors driving. It’s a profoundly bad idea to look at a set of statistics like this and say “oh, the ones that look inconvenient to me were probably doing something unsafe, they don’t count”.
On the other hand, these statistics don’t take health benefits from being physically active into account, which could potentially give ammunition for a much stronger critique—though given ike’s comments, I’m not sure it’d be a valid critique in the context of Jewish law.
But there are more and less safe driving habits, too, and I’ll bet the variance is about as high.
I bet less. Yes, you can practice defensive driving, but if you’re on the road in the traffic there is only so much you can do to avoid the idiot who is both in a hurry and needs to send that text message right now. You don’t have much control over external factors. But in swimming you often do—it’s pretty hard to drown if you are swimming in a pool with others watching.
it’s to compare their average dangers as they’re actually practiced
Yes. Therefore if you know you practice in way that’s different from the average, the probabilities change for you.
Yes, you can practice defensive driving, but if you’re on the road in the traffic there is only so much you can do to avoid the idiot who is both in a hurry and needs to send that text message right now.
I wasn’t thinking about defensive driving, I was thinking of driving thirty miles over the limit while not wearing a seat belt and texting your girlfriend about the awesome fight you just saw in the pub.
In pretty much any activity you can asymptotically drive your chance of surviving towards zero if you set your mind to it :-/
If we are talking about variance, the lower safety bound is often in approximately the same place, but the upper safety bound (as well as the center of the distribution) varies.
Yes, but if you’re going climbing you can choose to go the climbing gym and be absolutely safe from the avalanches in the Himalayas. However if you’re going driving on public roads, you cannot make yourself absolutely safe from drunk drivers.
You can make your climbing safer than you can make you driving.
That’s what makes climbing higher variance than driving.
You can make your climbing safer than you can make you driving.
You can make your climbing safer than summiting K2 would be, certainly. But enough safer to overcome those one and a half orders of magnitude of difference in the average? I haven’t actually seen any numbers on this, but that seems optimistic to me.
I’ll have to look at the methodology to believe that one and a half orders of magnitude, but regardless of that yes, you can make your climbing safer.
For example, you can do bouldering on technical routes which are all about agility and finger/arm strength. These routes rarely go more than 10 feet above thick mats—since you’re not belayed, you’re expected to just jump down when/if you run into trouble. Twist you ankle, sure, possible. Die—not very likely.
They’re a couple orders of magnitude riskier, actually. It’s tricky to make a direct comparison because the risk of driving is usually expressed over distance traveled, while sports is usually measured over number of sessions, but if we assume a typical day’s driving is about 50 miles (80 km), then we’re looking at 0.1 micromorts per session, as opposed to 17 for swimming or 3.1 for rock climbing.
(I’m not totally sure I trust that swimming estimate. The one for rock climbing aligns with my intuition, although there’s a lot of variance within the sport—bouldering is comparatively safe, while attempting the world’s highest peaks is absurdly risky by sports standards. I did know one guy who died in a shallow-water blackout and none who died climbing, for whatever that’s worth.)
[ETA: The estimate for swimming turns out to be bogus. See below.]
The link you gave puts car deaths above swimming in the second diagram. It doesn’t say that the sporting numbers are measured by session. (Except for the BASE jumping, hang-gliding, scuba diving, canoeing, or rock climbing). My own research (the first three links from Googling “risk of car accident death”) puts car accidents consistently higher than swimming deaths.
http://www.livescience.com/3780-odds-dying.html: 1-in-100 lifetime car death , 1-in-8,942 swimming death.
http://www.riskcomm.com/visualaids/riskscale/datasources.php: 1 in 17,625 one year car occupant death rate (based on 2002 data), 1 in 83,534 one year drowning death overall, 1 in 452,738 one year drowning death in swimming pool
http://well.blogs.nytimes.com/2007/10/31/how-scared-should-we-be/?_php=true&_type=blogs&_r=0: 1 in 84 lifetime car deaths, 1 in 1,134 swimming deaths.
I believe that’s because people drive much more than they swim, and the risk communication scale uses, say, your second numbers, and the comparison the link author gave converted that from annual to per-act.
I was trying to show that the swimming estimate wasn’t per session. 1 in 56,587 is close enough to 1 in 83,534 that they’re probably measuring the same thing, namely yearly deaths, in which case (assuming most swimmers swim more than 20 times a year, which I think is reasonable), the per-session risk for driving is more than that for swimming.
You’re right, it’s not per session—but it isn’t per year either. On closer examination it looks like they’re calculating the risk of death over the ten years surveyed (unless the 31 deaths reported are annualized, which I don’t think they are), which is an absolutely terrible bottom line—but fine, it makes the annual risk of death 1 in 566,000. I also notice that the population estimate is identical to that for running and cycling, so it’s probably some sort of very crude estimate of Germans involved in sports. Ugh. At least the climbing stats look more reliable.
Incidentally, an annual risk of death of 1 in 566,000 and a hundred sessions per year (two a week with time off for good behavior) gives us a per-act risk of 0.017 micromorts, about equal to driving four miles in a car.
It’s definitely not the chance of death in a year of swimming. My link already gives us all the numbers we need to calculate that—the number of deaths overall, the number of years being examined, and an estimate of the population involved—and it comes out to a chance of 1 in 5,658. (1,754,182 people / (31 deaths / 10 years).)
This conveniently lets us infer how they’re probably calculating the risk—it looks like they’re assuming one hundred sessions per year (or about two a week; fair enough) and doing a per-session estimate based on that. I also notice that the population estimate is identical to that for cycling and running, so it’s probably some sort of estimate of the number of people in Germany involved in an arbitrary popular sport. Cruder than I’d like, but I was only shooting for an estimate good to within an order of magnitude.
Those numbers look like general population numbers (and since it looks like a lot of drowning deaths are due to ineptitude, it seems unclear to me whether the yearly risk for frequent swimmers is higher or lower than for non-frequent swimmers). Instead of ‘all drowning,’ the 1 in 83,534 number, one should probably use the ‘in swimming pool’ number, which is 1 in 452,738.
I’m not sure I trust these estimates—or, rather, I don’t think I find them useful. The main problem is that the probabilities involved are all strongly conditional.
Consider swimming in a hotel swimming pool with a lifeguard watching and long-distance swimming alone in the ocean. Both are “swimming” but these two activities are radically different from the risk perspective. Similarly, you can do “climbing” in the climbing gym and you can do “climbing” in the Himalayas.
Sure, there’s a lot of variance involved. But there are more and less safe driving habits, too, and I’ll bet the variance is about as high. The point isn’t to demonstrate that one practice is under all conditions more or less safe than another, it’s to compare their average dangers as they’re actually practiced. And that clearly favors driving. It’s a profoundly bad idea to look at a set of statistics like this and say “oh, the ones that look inconvenient to me were probably doing something unsafe, they don’t count”.
On the other hand, these statistics don’t take health benefits from being physically active into account, which could potentially give ammunition for a much stronger critique—though given ike’s comments, I’m not sure it’d be a valid critique in the context of Jewish law.
I bet less. Yes, you can practice defensive driving, but if you’re on the road in the traffic there is only so much you can do to avoid the idiot who is both in a hurry and needs to send that text message right now. You don’t have much control over external factors. But in swimming you often do—it’s pretty hard to drown if you are swimming in a pool with others watching.
Yes. Therefore if you know you practice in way that’s different from the average, the probabilities change for you.
I wasn’t thinking about defensive driving, I was thinking of driving thirty miles over the limit while not wearing a seat belt and texting your girlfriend about the awesome fight you just saw in the pub.
In pretty much any activity you can asymptotically drive your chance of surviving towards zero if you set your mind to it :-/
If we are talking about variance, the lower safety bound is often in approximately the same place, but the upper safety bound (as well as the center of the distribution) varies.
I’ll bet there are more idiot drunks on the road than there are Himalayan mountaineers, even proportionally.
Yes, but if you’re going climbing you can choose to go the climbing gym and be absolutely safe from the avalanches in the Himalayas. However if you’re going driving on public roads, you cannot make yourself absolutely safe from drunk drivers.
You can make your climbing safer than you can make you driving.
That’s what makes climbing higher variance than driving.
You can make your climbing safer than summiting K2 would be, certainly. But enough safer to overcome those one and a half orders of magnitude of difference in the average? I haven’t actually seen any numbers on this, but that seems optimistic to me.
I’ll have to look at the methodology to believe that one and a half orders of magnitude, but regardless of that yes, you can make your climbing safer.
For example, you can do bouldering on technical routes which are all about agility and finger/arm strength. These routes rarely go more than 10 feet above thick mats—since you’re not belayed, you’re expected to just jump down when/if you run into trouble. Twist you ankle, sure, possible. Die—not very likely.
Yes, I mentioned bouldering in my original post.