« Which 20th century classic of American conservative political thought has held up best? | Main | Forward markets in everything, restaurant edition »
Suicide fact of the day
Glen Whitman reports:
I went back to the original data source (imagine that!) and found that the stereotype is dead wrong: suicide rates are notably lower for teenagers than adults...Suicide rates do rise throughout the teen years, but they plateau at about age 20 and remain flat throughout the years 20 to 65. Then they jump again for the 65+ demographic.
In case you're wondering, teen suicide rates have not been rising, either. They've been in decline since the late 1980s.
Yet these teens still take the most risks.
Posted by Tyler Cowen on February 28, 2008 at 11:23 AM in Data Source | Permalink
Comments
Any discussion about teen suicide needs to take sexual orientation into account. The academic literature is teeming with studies showing that gay teens, especially males, are far more likely to ideate about suicide, perhaps by a factor of ten, and also more likely to attempt suicide.
One would like to think that the decline in teen suicide rates that you mention is at least partially due to the gay rights movement and our society's increasing intolerance toward the intolerant.
Posted by: KipEsquire at Feb 28, 2008 11:39:12 AM
I wonder what the figures look like for suicide attempts. I'm not theorizing that it would reverse Glen's reported outcome, but I imagine when adults decide to kill themselves, it is more often "successful."
Posted by: Bob Murphy at Feb 28, 2008 11:44:53 AM
Tyler -- there is a HUGE difference between taking risks (entrepreneur! downhill skier!) and committing suicide (life is hopeless!). I should hope that the gradient would steepen with age, as the troubles of life settled in, but we need the risk taking exuberance of youth to propel innovation and excitement (let alone new babies :). Risk good / Suicide bad.
Posted by: David Zetland at Feb 28, 2008 12:06:38 PM
Dying as a result of risky behavior is recklessness. Not suicide. That is surely accounted for in the statistic?
Unless they die in an accident as a result of extremely risky activities that nobody knew they planned to attempt. That may be misclassified.
Posted by: HeShootsAndScores at Feb 28, 2008 1:14:22 PM
It's my understanding that these statistics underreport suicide rates. A teen who shoots himself in the head dies from suicide. A teen who deliberately drives his car into a retaining wall dies from an auto accident.
Posted by: Ted Craig at Feb 28, 2008 1:25:46 PM
> Suicide rates do rise throughout the teen years,
Well duh... what do you suppose the suicide rate for children 11 and under is? Practically nil?
Posted by: chuck at Feb 28, 2008 1:40:28 PM
Ted - I thought the same thing, but if anything, I would expect undercounting of suicides to be higher for adults than teens. Adults who commit suicide usually/often invalidate their life insurance policies, and therefore there's an incentive to conceal the nature of the suicide, moreso than for teens where that particular motivation is absent. Teen suicide may be more likely to commit public suicides, if it is motivated by mistreatment by their caregiver (eg, parent) or a scorned lover. When it is either, isn't the motivation to punish the person, which requires that the suicide be easy to interpret?
Tyler - Theoretically, suicide probabilities rise when the expected value of future consumption falls below zero. If anything, the fact that teens take risks seems to be that they are highly optimistic, not pessimistic, about their futures.
Kip - interesting hypothesis. You could use the Census 5% for 1990 and 2000, and following Dan Black's papers on the locations of gay couples, try to build a measure of gay cities. Then you could see if it is falling in those cities. If there was a correlation with even the states, that would be interesting. One thing is interesting is that if you look at the time series through the link, the fall in suicides starts in the early 1990s, which is when the Donahue-Levitt abortion-crime hypothesis predicts declines in crime. I wonder if you did a triple difference using the youth in the early repeal states and compared them with individuals born just before repeal in those five states, then compared them with the Roe states, whether there is any noticeable decrease in suicide probabilities for the abortion treated states. I suspect that there is not, but the time series is at least consistent with that broad trend.
Posted by: jason voorhees at Feb 28, 2008 1:49:51 PM
Clearly there's a disconnection between the general perception of the frequency of teen suicide and the reality. I think it must be associated with the fact that almost all teens are in school. They're in a daily social environment where they're observed to some degree (even if it's just attendance). This means that when a teen commits suicide, people are aware of it, even if the teen has few social connections before they die. When you're in high school, everyone hears about every suicide...the news spreads like wildfire. Once you're an adult, it's entirely possible that you live your life almost entirely unobserved and its possible to kill yourself without making a ripple.
Posted by: ramster at Feb 28, 2008 2:01:11 PM
The misconception is so much simpler than all this. It's just a matter of intuitive and incorrect risk assessment.
I worry that my parent may soon die of a heart attack or cancer because in their age group that is the leading cause of death. I worry that my teenage sister will die of a car accident or of suicide, because they are the leading cause of death in her age group. However, for some irrationnal reason, I worry about my parent' life and my sister's life about equally.
Posted by: Martin at Feb 28, 2008 3:38:13 PM
A large fraction of adult male suicides are men with terminal diseases who are shortening the end by a few months. Twenty years ago, Time magazine published the pictures of everybody in America who had died by gunshot in one week. An unexpected percentage were old men who were dying in pain.
The tragedy level of teen suicide on a per capita basis is much higher.
Posted by: Steve Sailer at Feb 28, 2008 5:01:08 PM
There's a simple explanation for the perception of high suicide rates for teens... we don't expect them to die, and it's one of the most common causes. So it sticks in our minds.
For white males 15-19, accident rates are 47 per 100,000; suicide is 15; homicide is 10. For black teens males, it is accident 38; suicide 11; homicide 70.
That last little stat is particularly unfortunate.
Posted by: Bob Knaus at Feb 28, 2008 5:33:21 PM
I thought several years ago Tyler had linked to a paper on the "Economic Theory of Suicide." If I recall the conclusions of that paper would fit very nicely with this data.
Posted by: Ernest at Feb 28, 2008 5:56:22 PM
Yes, here it is but it requires payment:
http://links.jstor.org/sici?sici=0022-3808%28197401%2F02%2982%3A1%3C83%3AAETOS%3E2.0.CO%3B2-E
Posted by: Ernest at Feb 28, 2008 6:02:26 PM
Interestingly, I would imagine that more "feminine" older males would have lower suicide rates, because they would be better at maintaining social relationships.
In modern American society men seem particularly poor at maintaining social relationships based on care and affection -- as opposed to work, sport, and accomplishment.
A health problem that prevent the male from being productive will usually also sever most of his social ties, as well as his source of status.
The desire not to be pitied, coupled with poor affirmation skills -- mean that an ill, unproductive male both gets and gives little in social relationships.
Most people, both men and women, prefer to discuss their problems with women.
Outside of immediate family, most men have social relationships based on being productive, as opposed to providing emotional support.
By my theory men with only sons would have a higher suicide rate than men with daughters, because the daughters could provide emotional support, while the sons likely could not.
In my experience, older unhealthy men often live in severe social isolation. Extremely few people have any reason to contact them. This of course compounds the problem of monitoring any chronic diseases.
Basically, nobody wants to talk to old, sick, cranky men. It's depressing. Old women often retain some nurturing skills that make caring for them rewarding.
It's a tough cycle, since being old, sick, and in pain makes people cranky -- which makes healthy people avoid them like the plague.
Posted by: jim at Feb 28, 2008 7:04:57 PM
Like murder, suicide can be thought of as having different "degrees".
If reckless behaviour leading to another's death is "third degree murder" or "manslaughter" then what it reckless behaviour leading to ones own death?
An example is drug overdoses. A person who abuses drugs does so because they want to forget or block out their own (unhappy) life. That this may kill them is not enough to stop them, it may even add to the attraction. Sometimes they do die. Is this death accidental? Or suicide? Or halfway in between?
A lot of bad, [drunk] driving would also fit this pattern.
Posted by: doctorpat at Feb 28, 2008 9:41:42 PM
What if certain people have a higher propensity to commit suicide than others? This could be either genetic or due to circumstances. Preferences are strange things.
It would seem that these people would die out sooner than others. So, as a population ages, the suicide rate would naturally decrease. At age 25, those most predisposed toward suicide have already committed suicide.
Then, at 65+, people start committing suicide because the net present value of the remainder of their life is lower than ever and rapidly decreasing.
Uh... I'm turning 22 in April...
Posted by: Paul R. Dorasil at Feb 29, 2008 1:21:45 AM
The amazing thing to me is the difference in suicide rate between rural and urban areas. In the U.S., the suicide rate in the most rural counties is twice what it is in the most urban counties. Is it loneliness or just better access to firearms?
Posted by: John S. at Feb 29, 2008 6:57:56 AM
John S. - or is it simply heterogeneity. People choose where they live to some degree. Perhaps the correlation isn't causal, but due to just differences in the kinds of people who choose to live in each. It could be that the people who have the highest expectation of their futures leave the town for the city, for instance.
Posted by: jason voorhees at Feb 29, 2008 10:59:24 AM
I think the increase of indulgence in other vices helps curtail suicide in a weird way. My assumption is that teens have more easy access to drugs/alcohol which actually might end up reducing suicide rates. The effect of this is then increased death rates over a period...(40s-50s)
Posted by: Arun at Feb 29, 2008 11:42:51 AM
How does a teenager have *more* access to drugs/alcohol than 40-50 year olds?
Posted by: Paul R. Dorasil at Feb 29, 2008 4:21:23 PM
Could there be an inverse correlation between risk-taking behaviour and suicidal tendency?
Posted by: Geoff Hamilton at Feb 29, 2008 5:45:44 PM
The suicide rate is highest, and rising, for middle aged men. Somebody sent me an article on this last week -- on my birthday (thanks, sis).
http://www.nytimes.com/2008/02/19/us/19suicide.html?_r=1&st=cse&sq=suicide+middle+aged+men&scp=1&oref=slogin
I have some thought that this might be connected to the decline in testosterone among men at this age (a decline which is unevenly distributed among the population). There are a lot of other theories possible though.
But nobody much cares. There are a lot of articles about depression/suicide among the elderly (Jane Brody seems to write about this every 4 months or so). Teen suicide gets a lot of press (a gigantic fear among parents). Middle aged men are supposed to be able to take care of themselves.
Posted by: ZBicyclist at Mar 1, 2008 10:16:43 AM
Lesser responsibilities and socially more acceptable when you are 18 as opposed to 45 with two kids....no?
Posted by: Arun at Mar 3, 2008 2:12:14 PM
