“of all would-be jumpers who were thwarted from leaping off the Golden Gate between 1937 and 1971 — an astonishing 515 individuals in all — he painstakingly culled death-certificate records to see how many had subsequently “completed.” His report, “Where Are They Now?” remains a landmark in the study of suicide, for what he found was that just 6 percent of those pulled off the bridge went on to kill themselves. Even allowing for suicides that might have been mislabeled as accidents only raised the total to 10 percent.”
In other words, if you ever think you want to kill yourself, there’s a 90% chance you’re wrong. Behave accordingly.
Well, yes, it just establishes a prior. But a remarkably hard prior to update, don’t you think? “I’m probably in worse shape than all those people who tried to jump off the Golden Gate Bridge” would demand some exceptional new information.
All this data says is that between 90% and 94% of people who are convinced not to jump did not go on to successfully commit suicide at a later date. It would be a big mistake to assume that whether or not you would come to regret your choice is 100% independent of whether or not you can be convinced not to jump and that therefore the fraction of people who came to regret commiting suicide is the same as the fraction who would have come to regret commiting suicide if they had failed their attempt.
“Apprehended” isn’t synonymous with “convinced not to jump”, but there does seem to be a sampling bias here, yes. (And can I say how refreshing it is to hear someone point that out and not be ignorantly insulted for it by dozens of people? Hyperlink to a “More Wrong” website omitted in the name of internet civility, but take my word for it that I’m describing an actual event.)
I think even “convinced not to jump” wouldn’t necessarily change the decision calculus here, though. To the extent there is a selection bias it’s because some subset of suicidal people behaved in ways which caused them to avoid opportunities to have their minds changed. That’s so irrational you could practically write a book about it.
One old study about one bridge is not the whole body of evidence regarding suicide, either. Read a few more bits from just that one news article.
Suicide rates reduced by a third in Britain merely because one easy method became unavailable? In other words, a large minority of would-be suicides didn’t even need to be convinced by someone else, they just needed less time to convince themselves than it would have taken them to find a slightly less convenient way of killing themselves. Even “very slightly less convenient” can provide enough time: 4 bridge jumpers per year were all deterred by one new barrier at the Ellington bridge, the local suicide rate went down by 4 jumpers per year, and the suicide rate at the unprotected, easily visible neighboring bridge only went up by 0.3 per year?
I personally wouldn’t have predicted any of this, but I don’t think there’s any major flaws in the data now that I’ve seen it. The biggest selection bias here may be one for those of us who naturally try to predict how people will rationally respond to changing incentives: applying such predictions to a tiny fraction of the population which has already self-selected for irrationality is not going to work well.
Suicide rates reduced by a third in Britain merely because one easy method became unavailable? In other words, a large minority of would-be suicides didn’t even need to be convinced by someone else, they just needed less time to convince themselves than it would have taken them to find a slightly less convenient way of killing themselves.
I think this quote is objectively accurate:
In other words, if you ever think you want to kill yourself, there’s a 90% chance you’re wrong. Behave accordingly.
That isn’t what the quote tells you. It is evidence that you could be wrong but certainly doesn’t make you 90% likely to be wrong.
Well, yes, it just establishes a prior. But a remarkably hard prior to update, don’t you think? “I’m probably in worse shape than all those people who tried to jump off the Golden Gate Bridge” would demand some exceptional new information.
All this data says is that between 90% and 94% of people who are convinced not to jump did not go on to successfully commit suicide at a later date. It would be a big mistake to assume that whether or not you would come to regret your choice is 100% independent of whether or not you can be convinced not to jump and that therefore the fraction of people who came to regret commiting suicide is the same as the fraction who would have come to regret commiting suicide if they had failed their attempt.
“Apprehended” isn’t synonymous with “convinced not to jump”, but there does seem to be a sampling bias here, yes. (And can I say how refreshing it is to hear someone point that out and not be ignorantly insulted for it by dozens of people? Hyperlink to a “More Wrong” website omitted in the name of internet civility, but take my word for it that I’m describing an actual event.)
I think even “convinced not to jump” wouldn’t necessarily change the decision calculus here, though. To the extent there is a selection bias it’s because some subset of suicidal people behaved in ways which caused them to avoid opportunities to have their minds changed. That’s so irrational you could practically write a book about it.
One old study about one bridge is not the whole body of evidence regarding suicide, either. Read a few more bits from just that one news article.
Suicide rates reduced by a third in Britain merely because one easy method became unavailable? In other words, a large minority of would-be suicides didn’t even need to be convinced by someone else, they just needed less time to convince themselves than it would have taken them to find a slightly less convenient way of killing themselves. Even “very slightly less convenient” can provide enough time: 4 bridge jumpers per year were all deterred by one new barrier at the Ellington bridge, the local suicide rate went down by 4 jumpers per year, and the suicide rate at the unprotected, easily visible neighboring bridge only went up by 0.3 per year?
I personally wouldn’t have predicted any of this, but I don’t think there’s any major flaws in the data now that I’ve seen it. The biggest selection bias here may be one for those of us who naturally try to predict how people will rationally respond to changing incentives: applying such predictions to a tiny fraction of the population which has already self-selected for irrationality is not going to work well.
Not that I don’t think that most people who plan to kill themselves will tend to think better of it as time passes, but it’s a mistake to assume that trivial inconveniences only prevent people from doing things they don’t really want or believe are good for them.
If you ever think you want to kill yourself, there’s a 90% percent chance that, either you’re wrong, or you will be after surviving the attempt.