Firearms and Suicides in US States

Suicides outnumber homicides in the United States by 3:1. (In 2010 there were 38,364 suicides and 12, 996 homicides.) Lots of studies have investigated the relationship between firearms and homicide but the potential for reverse causality makes this a difficult problem. More homicides in a region, for example, might cause an increase in gun ownership so a positive correlation between guns and homicide doesn’t tell you which is cause and which is effect. Reverse causality is less of a problem for understanding the guns to suicide link because it’s less likely that a rash of suicides would encourage gun ownership.

In my latest paper, Firearms and Suicides in US States, (written with the excellent Justin Briggs) we examine the easier question, what is the relationship between firearms and suicide? Using a variety of techniques and data we estimate that a 1 percentage point increase in the household gun ownership rate leads to a .5 to .9% increase in suicides.* (n.b. slight change in language from earlier version for clarity.)

Even if one thinks that suicides don’t cause gun ownership one might imagine that they are correlated due say to a third factor such as social anomie. We have an interesting test of this in the paper. If suicides and gun ownership were being driven by a third factor we would expect gun ownership to be correlated with all suicides not just gun-suicide. What we find, however, is that an increase in gun ownership decrease non-gun suicide. From an economics perspective this makes perfect sense. As gun ownership increases, the cost of gun-suicide falls because guns are easier to access and as the cost of gun-suicide falls there is substitution away from non-gun suicide.

Put differently, when gun ownership decreases other methods of suicide increase. Substitution among methods is not perfect, however, so when gun ownership decreases we see a big decrease in gun-suicide and a substantial but less than fully compensating increase in non-gun suicide so a net decrease in the number of suicides.

Our econometric results are consistent with the literature on suicide which finds that suicide is often a rash and impulsive decision–most people who try but fail to commit suicide do not recommit at a later date–as a result, small increases in the cost of suicide can dissuade people long enough so that they never do commit suicide.

The results in the paper appear to be robust but the data on gun ownership is frustratingly sparse due to political considerations.


"Even if one thinks that suicides don’t cause gun ownership..."

I'm in that camp for the simple reason that I think it's hard to buy a gun after you're dead, although I suspect a case might be made for contemplated suicides causing gun ownership.

I think Tabarrok is either shooting from the hip, or he's got his alleged cause and effect backwards.

That's the best you got? Clearly he means gun ownership in the community. He even says in the earlier paragraph "More homicides in a region [...]" so it's clear that he means that rising suicide rates probably don't cause surviving members of the community to buy guns, not like they may be tempted to do when the homicide rates rise.

Sorry JWeinberg, even if that is what Alex means, I think I'd have to put it in the category of "duh".

Vivian, I think that science happens when "duh" stops.

Instead of assuming what you think you know, you look at the possibilities and what the data tell you about the possibilities.

When dealing with the question of "what causation is implied by this causation," it is wise to consider that X causes Y, that Y causes X, and that X and Y are caused by a third factor Z.

Alex is behaving appropriately.

You are behaving like one of the many internet people who feel compelled to reply to any real science "duh, I didn't need a study to tell me that!" Common sense is often wrong, and the things we "know" to be so, are often not so. That's why the serious people in the room do real science.

The old "duh, it's common sense" line.

That's the great thing about social science empirical work. It always just discovers commons sense like "birds of a feather flock together" and "opposites attract." Oh wait....

JWeinberg is correct.

"science happens when “duh” stops." - that's brilliant, mrpinto. Can I attribute it to Albert Einstein and spread it over the internet? hehe

In the US, who decides on cause of death? I ask because my pal The Retired Epidemiologist told me about an attempt to compare causes of death in Britain and Italy: they were particularly interested in suicide. They learnt that in Italy cause of death might be determined by any of (i) doctor, or (ii) nurse, or (iii) family of the deceased: they abandoned the study.

I think we can count on gunshot deaths being actively and professionally investigated. (No doubt some murders are successfully staged as suicides, but I wouldn't think to a level affecting statistics.)

I think the relevant question isn't murder vs. suicide, but accident vs. suicide. The family may prefer thinking it was an accident (or at least having that be the public view), for example.

So you're saying that gun suicides are harder to hide as suicides than other means of killing oneself which might mean some or all of the apparent decrease in suicide could be due to reclassification rather than fewer suicides?

It's possible, but I'm not certain it's harder to hide a gun suicide as an accident than most other forms of normally effective suicide (an "accident" while cleaning a gun seems no more or less tough to pass off as an accident as "accidentally" falling off a tall building for instance). And beyond that insurance companies are likely to want to have an accurate cause of death as that could activate as suicide clause allowing them to deny a death benefit. My intuition on this is thus that the effect, if it exists, is likely to be small.

Unless the policy is new (less than two years old), suicide is generally covered by life insurance.

Right, I wasn't thinking about the non-gun suicides, but I have often wondered how many people who are going to suicide set out the gun-cleaning supplies, to make a "cleaning accident."

Though in order to invalidate the data you need to establish that 10-50% of non-gun suicides are misclassified as accidents (plus whatever percentage of gun suicides are also misclassified), as well as establish that Alex didn't already take that into account in his calculations (I haven't read the actual paper yet).

As an aside I find the hypothesis that increased guns leads to increased suicides to be obvious. It's basic economics, you make an activity easier and you'll get more of it.

We also know from studies of which deaths inspire most fear that "visualized pain and suffering" drive it. I'd think that guns, with less visualized pain and suffering, would be more attractive deaths.

Trying again with the link

Catholic countries have different standards/protocols, re: suicides ... because it is a sin.

Not so in Protestant countries, where it is (generally) frowned upon but not elevated to sin status. This almost certainly skews the data.

Catholics and Protestants agree that suicide is a sin. The disagreement is over whether it is forgivable.

Nitpicking aside, though, it is true that cultural attitudes toward suicide skew the underlying statistics, making inter-country comparisons quite difficult.

We did a back-of-the-envelope analysis of U.S. versus Canadian suicide rates for for the years from 2000 through 2007. The analysis is applicable here because Canada has much more restrictive gun ownership policies than the U.S. does, which provides a pretty direct way to measure the impact of firearms upon the rate of suicide.

What we found is that firearm availability has very little effect over the incidence of suicides, as the suicide rate in both nations is nearly identical and that the major difference between them is the selected method of suicide. Canadians who commit suicide do indeed substitute hanging/suffocation, poisoning and other methods for firearms.

This assumes that Canada is a good comparator for the United States, which it is likely not. It is likely that suicide rates increase along with distance from the Equator:

Note also that Alaska is consistently among the highest-suicide states in the US.

Also by county:

Note that this pattern is clear - where there is white rural poverty and high gun ownerships, there is suicide.

In another thread we talk about population genetics, and how that is different than gross racism. Interestingly as someone with 25% Icelandic heritage (by way of Canada), I might, just might, have inherited protection from this (wikipedia): "The study's authors suggested that propensity for SAD may differ due to some genetic factor within the Icelandic population. A study of Canadians of wholly Icelandic descent also showed low levels of SAD."

I expect that there was fairly vigorous selection against susceptibility to SAD among the Icelandic population over the last thousand years.

Does natural selection work over 1,000 years?

Absolutely, if the selection pressure is strong enough.

Selection pressure can work over 1,000 years if the selection pressure is strong enough (many disease related selection events happen in a matter of decades or less), but in the case of Icelanders it wouldn't have to. Iceland was settled from places which were already at high latitude that involved similar selective pressures although less intense ones with more dilution from in migration from more Southerly areas.

Far northern Europe has been in relative population genetic stasis (apart from local random population genetic drift and selection effects) for roughly 3500 years (i.e. since the late Bronze Age).

The relationship holds when it is limited to just the states and provinces that directly border each other.

So many confounding factors in that analysis; the proper natural experiment is for two similar cities in different jurisdictions, comparing say Seattle to Vancouver, or Buffalo to a city in Ontario. I mean, come on, there should be an obvious confounding factor when you include populations that partially live in areas with harsh winters and low light.

"Does natural selection work over 1,000 years?" What, you think it goes on strike?

I am reading this with sadness. A 16 year old kid on my street took his own life last week, with a rifle.
I am all for gun control. But seems to me the politicians are pushing it for the wrong reason. Why the focus is on the mass shooting instead of suicides?

Because you convince the laity of the justness of your cause based on the most empathetic scenario, not the average or most common. This is known as Politics.

Also one action is a voluntary expression and the other is an act of violence upon others.

When my brother's best friend committed suicide with his dad's gun, a lot of other parents in the community either got rid of their guns. That's just anecdotal data, but it got me thinking about the possibility of a causal effect from suicide to gun ownership rates. While homicides may increase ownership rates, is it possible that suicides (by gun) decreases it?

I got rid of my gun years ago for fear that my children would find it and someone would be hurt by accident.

That is a great question. And Joe Smith's comment suggests that "number of children" should somehow be included as a control (children in the house? children in the community?).

Because it's quite a different calculus to say "person X cannot have a gun because person Y might get one and consciously hurt himself with it." There are other ways to stop person Y.

I fully believe access to guns increases suicides, BTW. People have short-term desires to commit suicide but only follow through with it if they can do it quickly. If it's too hard to kill yourself in the gas-powered oven, people don't concoct other plans to kill themselves. They give up.

Or they cut themselves, something they can recover from. Not so (obviously) with a gunshot.

Obviously experience varies. I only have been close to two suicides. One, my father, started his car in the garage 46 years ago, and was dead before long. The other happened very recently on our street. A young man, 17, for unknown reasons, decided to kill himself. He drove more than 90 miles to the George Washington Bridge (NYC) and jumped.

I would dispute the "farther from the equator" theory given above. Japan is closer to the equator than Alaska or Canada. As for guns, people theorize that an attempt is more often successful. Compared to what? As for ease and speed of accomplishment, jumping in front of trains or 18 wheelers (at highway speed) is simple, accessible, and reliable, as are many self-poisoning methods. Suicide is about the mental process, not the means.

Suicide is a personal choice. Being killed by another is not.

What's the difference between making love and rape? Between going on a holiday and kidnapping? Between spending one's own money and theft?

Stop treating dissimilar things as if they were the same.

And this...
"most people who try but fail to commit suicide do not recommit at a later date–as a result, small increases in the cost of suicide can dissuade people long enough so that they never do commit suicide." non-sequitur too. Because people who TRY AND FAIL have a credible signal that they need help, which often enables them to a degree of rent-seeking that is not available to anyone who omits suicide. You might just as well pay everyone to stay alive.

The implication of your comment is you actually resent suicidal people because of the level of support they receive.

There's definitely a short term increase in support, but the one girl I know who attempted suicide basically got kicked out by her roommates because they didn't want to deal with the fall out. Suicide attempts can also cause people to lose the serious social connections they need.

I think one reason why failed attempts don't re-attempt is because a lot of them aren't intended to succeed, they're still incredibly dangerous and may succeed, but they were more a cry for help rather than a sincere attempt to die.

"The implication of your comment is you actually resent suicidal people because of the level of support they receive."

Actually, it isn't, and I don't resent suicidal people at all; they have the right to self-determination. I also happen to think that other people shouldn't be forced to pay for their extra support.

The point I was trying to make is that suicide attempts bring benefits to survivors that omissions of suicide don't bring. You're probably correct that there are also downsides. So the lack of re-committing could be interpreted as

a) fear of the repetition of these downsides
b) improvement of the condition because of extra support (which causes equivalent costs to others)
c) the temporary delay in suicidal efficacy, resulting in a stable preference for life
or some combination thereof.

Restricting method access assumes (only) c) is true, without providing evidence for it.

Mass shootings predominantly are suicides.

It's touching that the policy wonks want to infringe upon our Second Amendment rights to protect us from choosing the time of our own passing, while instituting death panels, ahem Independent Payment Advisory Boards, that will ensure that power properly resides in the hands of the wise and beneficent state.

Definitely agree. It's far better to let some faceless insurance bureaucrat decides who gets insurance and who can go die in a ditch rather than some government advisory board accountable to the Congress, the President, and the American people.

None of those things actually happen.

I'm sure folks in Britain couldn't imagine 60,000 people a year starving and dying of thirst on the Liverpool Care Pathway either.

The left-wing intelligentsia is in favor of bureaucrats making sagacious decisions to terminate the old and ill, but I doubt very much the masses of low-information voters who put them in power actually favor these policies.

Oh, but whatever - let's let these same wise government paper-pushers chip away at our gun ownership rights because some people might decide to express their autonomy and choose the time of the own deaths.

If my choices are to let a hapless government decide my health outcomes or a profit-driven group of middlemen whose incentives are to simply deny claims and dare you to sue, I'm gonna go with the former.

Americans actually have no experience, generally, with government medicine. Even when the government pays, the provider is private enterprise (subsidized by the government).

I'm American, but in Sweden (wife's farm, we're often there) I watched our tenant farmer of 40 year's standing die at age 70 because the triage decision was that the coronary bypass operation he needed was not high priority. He died waiting. Britain has similar types of experience. Government--run eventually means triage. The rich just by foreign insurance and get their bypass or transplant in another country. The "low-information voter" meme is not without some merit, though it sweeps both major US parties.

You can, and many states do, attack this problem with safes and gun locks. If teens don't have the combination, that will reduce teen suicides. (Per Alex's data that gun availability increases the actual rate of suicide.)

I dont know of any state that requires gun owners to have a gun safe and ive never met a gun owner that actually uses the gun locks that are standard with guns.

One of those links proves that gun safes are required by law in a state?

lazy mofo.

Not literally "gun safes for everyone," but North Carolina (the state with which I am most familiar with its gun laws) says that if someone under 18 shares your premises, it is a misdemeanor to leave a firearm in a state where an unsupervised minor can fire it. Which seems to be pretty much what john personna was talking about.

No it isnt really the same.

"You can, and many states do, attack this problem with safes and gun locks"
is not equal to
" it is a misdemeanor to leave a firearm in a state where an unsupervised minor can fire it"

Its not even really close.

The post reports that a 1% increase in gun ownership results in a .5% to .9% increase in suicides. Maybe it's just too early in the morning, but does this mean that every additional gun owned has a 50%-90% chance of being used in a suicide? I'm sure the relation isn't quite that simple, but the number is striking enough to make me suspicious of the econometrics.

Imagine a situation where there are 1,000,000 gun owners and 1,000 suicides.

OK, "too early in the morning" wins! Thanks.

As is always the case with these studies, sweeping generalizations are made to prove once and for all that correlation is, in fact, the same as causation. For starters, what's the relationship between say duct tape sales and people asphyxiating themselves with a running car in the garage? How about the sale of poison to the incidence of poisoning? I'm betting we find out that the tools used in a particular suicide are more common when the rate of that style of suicide is higher.

Then we have the issues of race, sex, age and class. Rich white guys are probably more fond of hanging than poor black women. Young people will prefer different methods than old people. There's a correlation for you. I bet you would find as suicide rises amongst the old, there is a greater likelihood of prescription meds in the home. Old people are more likely to OD rather than shoot themselves.

This is why I suspect the epitaph for humanity, at least western civilization, will include some mention of our inability to sort correlation from causation.,

Econometricians are easily capable of dealing with these problems, so I don't think they'll get a shout out in humanity's epitaph. Age, sex, race, and class can all be controlled for directly in the regression simply by adding extra variables. I'm sure data is available for all these classifications on people who have committed suicide. The paper is gated, but I wouldn't be surprised if they did exactly that.

The notion that areas with high suicide rates just have more guns because it makes suicide easier fails the sniff test. If this were the case, you would expect to see the excess guns purchased directly before a suicide attempt, which is probably not the case. High gun ownership areas have a lot of gun enthusiasts, hunters, or people worried about self defense.

I will say this once, and only once, because I am so god-damned tired of telling my students:

You don't *control* for anything in a regression model. You may (and should) try to adjust for the covariates, but you don't control for them.

Can you elaborate. All I'm reading from your comment is that there are multiple ways to kill oneself.

Sure, I'd be impressed if you analysed data to show suicides correlate with duct tape sales. Have you? Can you? A discovered correlation is different than a hypothetical one.

Correlation may not be causation but it sure is a good start.

It can be a good start or it can be a terrible start. If it leads you down the road to loopy arguments about water causing drownings, then it is not a great starting point, is it?

Look, if a town suddenly has a carpentry craze, would you be shocked to learn there is a spike in hammer sales? Would you argue that the availability of hammers causes carpentry? I hope not. If a town builds a bunch of new tall buildings and the incidence of suicide by sudden deceleration spikes up, are you going to claim tall buildings cause people to jump off them?

Yet, we're supposed to believe that the mere presence of metal shaped a certain way causes people to kill themselves.

To use your analogy, this would be akin to a carpentry craze where hammer sales spiked yet sales dipped for nails, glue, chisels, saws etc.

Did you read about the decrease in non-gun suicides?

I've caused enough commotion here so this will be my last post on the subject, but I'll answer your query. I'm never shocked when people use the better tool for the job. Therefore, I'm not shocked that more people are proficient at that job. Going back to my carpentry example, access to hammers would result in improved nail use compared to when people were forced to use rocks instead of hammers.

The conclusion I draw from this paper is that offing yourself with a gun is preferred and more effective than other forms of suicide. Take away the guns and the same people will go to other, less effective methods, thus resulting in a lower incidence of successful suicide.

Z has won this thread. Several times over. Relentlessly.

I would only add that gun suicides are quite clearly suicides and are labelled as such, whereas with non-gun suicides, not so much. When people drive into bridge abutments, or walk out onto the highway at 2am, or do five pounds of heroin, these are labelled accidents -- and not all of them are.

I'd therefore suggest that non-gun suicides are higher across the board then these numbers suggest.

This is ridiculous. We have statistics for both Canada and Britain. We know that if guns are made unavailable, people choose other means. Why? I would posit that a successful suicide is the result of determined pursuit. People don't kill themselves because the opportunity exists. If that were true, the railways and highways would be littered with bodies, and the hardware stores would be out of both types of drain cleaner.

If you want to decrease suicide, as opposed to merely handcuffing a person who seriously wishes to die, look at the social nexus of the person. Look for brutalization, financial disaster, alcoholism, serious group bullying, or imposition of other major shame, such as rape. It is these circumstances which are an evil, not suicide itself.

Alan H, yes, we have statistics, and they are in the same drawer with lies and damn lies. Do you really know what the statistics actually say? Example from the UK: "As events of undetermined intent are included in the National Statistics suicide definition, this ICD rule change could have potentially increased the number of deaths included in ONS suicide statistics."

"Correlation does not imply causation" isn't some magic phrase that negates all research you don't like. Every researcher knows that phrase. The challenge of research is how to identify causal effects given that problem. It's an issue that has been explored in depth for decades and decades. There are many known techniques for disentangling the problems. This post is, in fact, talking about some of those issues. I have not read this paper, but there are indeed papers where causality can be established. Any paper that manages to convincingly identify a causal relationship is in no way whatsoever making general claims that all correlation is the same as causation. To even think any paper would make the claim shows a profound misunderstanding of academic literature.

You're right; it is not a magic phrase and no one but you seems to think so. It is a statement of fact. All heroine addicts ate sugary cereal as kids. Only a lunatic would think sugary cereal leads to heroine addiction. This study simply shows that the tools of a task increase with demand. That's hardly revelatory. Of course, if he confined his study to hammers and sore thumbs, no one would care. Going after guns makes it perfect troll bait and a chance for all of the usual suspects to come out banging their pots and pans in the streets about gun control.

Certainly correlation does not prove causation, per se. But when you have a robust correlation between GUN OWNERSHIP and SUICIDE BY FIREARMS you at the very least have a very credible story, especially given that suicide is usually not an elaborately plotted event but an impulsive, rash decision made in a specific moment of deep depression and anxiety. It is very, very easy to see why, given that underlying fact pattern, if you make suicide-enabling machines more readily available, suicides may go up.

Adding adjectives in front of the word "correlation" does not make it more like the word "causation." Further, "a very credible story" is often, not always, called a myth. In your case, a convenient one that you are willing to swallow down without question.

I read this study and conclude that shooting yourself in the head is a far more effective method of suicide than swallowing pills or sticking your head in the oven.

"I read this study and conclude that shooting yourself in the head is a far more effective method of suicide than swallowing pills or sticking your head in the oven."

I don't get it. So you agree that in general, gun ownership is associated with higher *successful* suicide attempts?

Z: The last sentence is irresponsible.

In all but your final comment you ignored the fact that gun ownership correlated with an increase in TOTAL SUICIDE, not just suicide by gun.

It may be that an increase in sleeping pill sales causes an increase in sleeping pill suicides, but would they show a significant increase in total suicides?

As for hammers, it may be that more hammers significantly increases the rate of sore thumbs, and if we were really worried about sore thumbs we may they worry about hammer ownership.

As for it's relation to gun control, if you're this resistant to good evidence with limited confounders and a very clear story when it comes to guns and suicide it makes me extremely skeptical of your arguments on the trickier subject of the relationship between homicides and firearms.

The less widely known flip side to the "correlation does not imply causation" rubric is that correlation has a cause, at least if it is highly statistically significant.

If something that could very plausibly cause something else is strongly correlated, a convincing rebuttal calls for some alternative cause that could give rise to the correlation that is observed.

“Correlation does not imply causation” isn’t some magic phrase that negates all research you don’t like.

Thank you, thank you, thank you. Blockheaded internet commenters are constantly trotting out this phrase as if it's a bomb than annihilates empirical work. Yes, economists know about causality. Yes, we work very hard to deal with it as best we can in an imperfect world, with a variety of techniques which frankly internet commenters do not understand. Yes, we are better at economic research than you are. No, your attempted trenchant little criticism isn't new, isn't novel, likely isn't correct, and 9 times out of 10 has already been dealt with even if it isn't stupid. Feel free to try to have your non-causal correlation=causation regressions published if you think that's what economists do. You will fail.

Moving on, now let the adults have a conversation.


Your pet theories are just correlations, mine come with causation.

Interestingly even who love to spout that cliche will most often have no answer if probed as to what evidence exactly they'll accept as causation and not mere correlation.

Maybe if you bothered to ask such a question, the answer would not seem so elusive. Projecting your defects onto others does not reflect well on you, my friend. I think you are better than that.


If you have a good answer I'd love to hear it.

"Moving on, now let the adults have a conversation."

Pretty much what one comes to expect from smarmy fanatics trying to shout down inconvenient opinions.

Agree. As for the profession's presumed expertise with statistics, joined to their frequent failure to convey results with an adequate description of the limitations on their conclusions, the thing speaks for itself.

An elevated total suicide rate has to be tested against many many other contributing factors, including job dissatisfaction, real wages, social views on the sin/non-sin of suicide, and so forth. One study's output is interesting, but determinative of nothing.

And I would ask, from a different point of view, why are you so concerned with suicide, as opposed murder or negligent homicide? Why should we adjust social policy because of a minor change in overall suicide or most common method of suicide?

I think you are raising a numeric fallacy, that more specific numbers change the broader answer. If Alex's data shows that more guns mean more suicides, that's just what it shows, no matter what the subgroups of "race, sex, age and class" are.

But don't forget Simpson's Paradox.

I wish I could remember names of things. There is a well known test in behavioral economics where they ask people to predict the odds of earthquakes in California, and then earthquakes in California killing 10 people. People will have a tendency to put the odds higher in the second case. That's impossible, since it is more specific, but the fact that more specific things become "storified" makes them more plausible to the listener.

There is a name for this effect. Anybody?


Thank you, Mr. Fallacy.

Can you point to the literature on suicide you mention? I've read some of it (e.g., Thomas Joiner's work) and I don't think "rash" and "impulsive" are quite right (though I have read that most survivors don't later die by suicide). I'd suspect that guns make suicide easier not because they enable rash decisions, but because they make it easier to overcome the fear and difficulty of self-injury.

If you use Google Scholar and search for impulsiveness and suicide, you'll find plenty of journal articles about it.

There's also an interesting paper regarding guns, suicide, and the Israeli army. It's been a long time since I read it, but it had something to do with rates of gun-related suicides on weekends by soldiers. Israel decided to no longer send soldiers home for the weekend with their guns. Suicide rates over the weekends dropped. They also checked for a corresponding upticks in weekday suicides. I don't remember the details, like if they checked non-gun suicide, or how significant (or insignificant) any upticks in weekday suicides were.

One of the famous studies to establish the notion involved suicide prevention barriers erected on a bridge to prevent people from jumping off. Rather than increasing suicide rates at a neighboring bridge, suicides by jumping off bridges fell as the barrier was put up with no compensating increase at other bridge nearby. Citations at:

I assume from the description that you are only counting successful suicides?

Or even reported. If I take some pills and just get sick and throw them up, or drive off the road and only wreck my car, as opposed to putting the barrel of some gun in my mouth. There are lots of ways to attempt to kill oneself unsuccessfully but a gun attempt will either be successful or very obvious that it was an attempt.

Yes. It is very much like the recording of attempted and successful murders. An attempt to kill with one's fists or a club may well be entered as an aggravated assault. An attempt to defend with a pistol may well be entered as an attempted murder. Indeed, the rise of gun crime in South Chicago is largely the result (I have been told by two grandfathers) of skinny drug dealers being tired of getting beaten senseless by the types that formerly ran the business, brutal thugs good with their fists, feet, and clubs. I can't say the evolution is good or bad from the local's point of view.

The leading cause of suicide is having a high IQ. People with low IQs almost never commit suicide. Animals even more infrequently. IQ control now!!!!!

As is being a widower. Men over 65 comprise the highest group of suicides, and that increases if the man is a widower. Ban marriage and monogamy now!

More Russian brides!

Under Obamacare, Russian brides will have to be covered by your health insurance...

even if you're a woman

I mean the insurance company will have to buy you a mail order bride... doh

Well, no wonder blog comment quality consistently declines......

That's your opinion. Social life is a complicated mix of circumstance and choice. Instead of looking at actual social mores and distortions, a large group wish to look only at, well, guns. For instance, I note the recent and large increase in purchase of small-caliber pistols by Indian women, to avert rape, and I hail the choice. It may lead to a few extra suicides. It will lead to many many discouraged rapists. Increased gun ownership will display correlation with many outcomes, which outcomes will themselves have (in the Bayesian sense) many causal paths to occurrence.

Or just, you know, deal with the problems you can deal with.

Gated paper.

Does it account for both gun ownership and suicide method being self selected? Say an increase of 5000 guns within a state is associated with 1 additional suicide (or whatever the baseline numbers may be). How do we know that the guy who committed the extra suicide didn't buy the gun a few days before for that very purpose?

The authors can't possibly suggest that simple exposure to increased gun ownership within a state causes people to kill themselves who otherwise wouldn't. Or are they?

It seems like this is yet another weak gun-outcome correlation study, where the discussion ends with an appeal for more taxpayer funding to run more weak correlation studies (while of course invoking the NRA bogeyman).

In a broad population, if gun owners kill themselves more often than non gun owners, I think we can at least recognize it as a "cost of ownership."

And of course it might be good advice for Dad alone to keep the combination for the gun safe, and then to take that as his real responsibility, no suicide for himself.

Even if the guy committing the suicide bought the gun a day ago, that still proves the point, right? Perhaps a milder version.

Sorry if this was covered in the paper or in the comments already, but could it be that someone would go buy a gun if they wanted to shoot themselves?

From the paper, footnote#3 "The literature implicitly assumes that household ownership means enhanced access to a firearm although this assumption requires that there is some barrier to buying or obtaining a firearm in order to commit suicide. There are several reasons to think that although by no means insurmountable,the need to find and purchase a gun may be enough to deter in many cases. First, it takes time and effort to find and go to a fire arm dealer, choose a weapon, get a background check, and obtain it (possibly after a mandatory waiting period), and this may give time for reflection and re-evaluation. Second, for many the price of a firearm and ammunition is not trivial, starting at several hundred dollars. Third, a potentially suicidal individual may be paralyzed by oppressive thoughts that preclude seeking to purchase a firearm.Finally,there may be cognitive barriers that cause an individual to think obtaining and properly using one would be harder than it is. For there are to be some effect of current possession rates, any marginal barrier will suffice. Ownership is likely also correlated with other measures of ease of access."

Thanks for asking before ranting =)

Freakonomics has an outstanding 60 minute podcast on suicide:

Yes, guns are not the cause of suicide but they are closely related.

A comparison with cars and road traffic accidents deaths may explain the meaningful "correlation but not causation". The cause of the accident is never the car even when a mechanical failure is involved, mechanical failures come from bad engineering design or lack of proper maintenance. However main road death causes are driving intoxicated, overspeeding, driving and texting, old people with impaired senses, exhausted even if the cars are not the cause of death there are laws that apply to drivers and cars to reduce fatalities. You have to get a license to drive, there are controls and penalties for DIU drivers, annual mechanical checks for the car, drivers or certain age have to take exams to keep the license, cars are built today with seatbelts, airbags and a reinforced chassis to absorb crash energy, roads are designed to be safer under bad weather, etc. but, in the end you die in the car and people talk about car or transport safety.

The same may apply to gun safety. It's not about stop selling guns, it's about minimizing the reckless endangerment of people. Yes, there are background checks for gun sale in stores, but you can easely go to your local gun fair or Yes, gun owners are responsible adults, but there are "accidents" like these: Empathetic people say that the parents have enough with the burden of a dead child and a life of regret, not so empathetic people would named it child abuse. Also guns used in the suicide of non intoxicated adults are problem free. Is this good? It would be better if the suicide went to the gun shop instead of using a family or a friend's gun? It's open to discussion. The results that Tabarrok shows than gun ownership increases a little suicide rates should also matter.

I think that if people in the Deparment of Motor Vehicles named the program "car control" would also face lots of contempt, think about "gun safety" and even gun owners may support it.

Other comparison with cars, for now there's a liability insurance for your guns. Even if you are responsible and don't have the intent to harm other people with your car, it is still a deadly machine under certain conditions. It would be great if the same applied to guns. Insurance companies would be happy to screen gun owners records and even more happy to charge more the bigger the gun and charge more with DIU, substance abuse or domestic violence records. Also, relatives of people killed by a gun under insurance would happy to receive a compensation. If deaths are not 100% avoidable, at least make them pay big bucks when it happen.

Gun insurance has existed for decades with regards to accidents (try your homeowner, renter, or umbrella policy), but not intentional acts. Try getting a reasonably-priced home policy if you've had a few gun-related (or any other kind) of liability claims.

As for the cars vs. guns issue in general, cars are not protected under the Constitution in the same manner. The current state of protection for the right to bear arms may not be desirable, but it is a major difference. If the right to drive cars had the same mention, I think you'd find "strict scrutiny" being applied to laws about driving too.

This is a fantastic idea. What other insurance policies can't we scheme up? I think we should have internet comment feelers insurance, in the event that our comments offend someone's sensibilities. In fact, I find it astounding humans have been able to live as long as they have without insuring every aspect of their lives. This must stop; insurance companies need more cash!.

Maybe we should have web sites register with the DMV along with newspapers and book publishers. Funny how the people making the car - gun comparison never apply the same logic to the other amendments. Probably just a coincidence.

BTW, gun sales at gun shows or on-line are not exempt from background checks. That's a myth propagated by the usual suspects.

You should broaden your reading to beyond the New York Times' anti-gun screeds. Here's a good start:

"Insurance for guns" is just another scheme to put more hurdles in the path of lawful gun ownership. No insurance company issues policies that pay out for criminal actions. If someone steals your car and drives it into an accident, neither you nor your insurance company are liable to pay anything to anybody. If you intentionally cause an accident with your car, your insurance company will not pay.

It's just a scheme to make insurance people more money, IMHO.

If you carelessly entrust your car to a drunk without a license, you and your insurance company will be on the hook when someone is injured in a crash.

It is, in the states whose law I know, illegal to lend a handgun to someone except under very limited circumstances. For example, in PA you may lend it only to a person with a carry license, an LCF. If you lend your gun to someone visibly drunk, even one with an LCF, no provision of state law will save you from massive liability should your action contribute to harm.

I find the post (and the research) an excellent example of Drs. Cowen and Tabarrok's ability to get out of the ideological foxholes and let good empirical results lead us. Rather than falling into an Animal Farm parody, "Freedom good, regulation bad", they are willing to point out cases for which their ideological bents might not easily apply.

I found the last bit to be funny: " on gun ownership is frustratingly sparse due to political considerations. "

Let me ask you, Alex, If you were doing a study on croc-pots, would you also write the line " on croc-pot ownership is frustratingly sparse due to political considerations. " ?

No, because no one wants to study the effects of croc-pot ownership on anything. There's no demand or use for that data, so no one collects it.

On the other hand, many researchers want to study gun ownership.

I suspect that croc-pot manufacturers would be quite happy to share information on the subject if you were to speak to reps at a kitchen ware convention.

The problem I have with these comparisons is that Japan and South Korea, have generally two to three times the suicide rate that the United States has. People at this point tend to get lazy and point to cultural differences with out much more thought (IMHO opinion if someone has to wave the magic wand of cultural difference to explain a place they don't understand it well enough).

Overall, whatever substitution rate there is for other methods, I suspect it is strongly affected by ones social and media sphere. In South Korea for example, there have been several high profile suicides by jumping off of buildings or hanging, there are fantastical yet exaggerated stories that during the Asian financial crisis (the "IMF" crisis to them, because they refuse to take blame) someone would jump off off one of the Han River Bridges in Seoul every 15 minutes.... Other methods in the US have a lower substitution because they simply aren't as popularly depicted or discussed, they simply lack social momentum.

What about measurement error?

A lot of suicides are actually homicides. A lot.

And gun ownership data is also bad. And the error is correlated with casualties. That is, when there is a casualty involving guns, you discover more guns.

If you want to claim something reliable, your estimates should be robus to this.

I am the other Ricardo. We need to disambiguate. How should we do that?

For example, we could do something like: if the first word in the lead article in tomorrow's New York Times begins with a vowel, you get to stay Ricardo, and I have to rename myself; otherwise vice versa. How might we go about this?

"A lot of suicides are actually homicides. A lot."

Why would you claim that this is the case? Pretty much every death by firearm is investigated by law enforcement and a professional coroner or medical examiner. Their means of distinguishing homicides from suicides are quite scientific. I have never seen any study or credible account casting doubt on the reliability of these inquiries, which are admittedly imperfect. Notably, some small percentage of cases are not classified as either due to lack of evidence, which suggests the degree of coroner perceived uncertainty in the classification process.

Indirect evidence that suicides are not in fact homicides in large percentages is that the demographic profile of suicide victims and statistical commonalities of deaths classified as suicides differs greatly from the demographic profile of homicide victims and statistical commonalities of deaths classified as homicides.

The only place homicides are routinely misclassified as homicides is on television. (Note that you could also be arguing that many suicides are a result of social mistreatment that causes people to kill themselves, but I don't perceive that argument in the comment).

Is there any way to look at how long the people who committed suicide had owned their guns. It seems to me that you might find that someone who goes out and buys a gun for the first time in his life at 55 might be at substantial risk of suicide in the months immediately after the purchase.

I don't have that data with me, but there is definitely a correlation between the use of a gun in a suicide and the length of time a gun has been owned, and also with the number of guns owned. Relying on memory I will say that three months out, the gun is less likely to be used for suicide. There are better numbers available. There is also a time-delay for guns and homicide. Putting aside the pathos of aging and loneliness, I would guess (no numbers) that one gun in a collection is less likely to be used for suicide than a person's sole gun, especially if recently purchased.

I hope not a penny of Federal money can be connected to this research, or else Alex will find himself subject to investigation and subpoena by Issa's staff. And the NRA will likely be attacking Alex's employer to get him fired.

Data and reason on guns is forbidden as much as on climate change because it is challenges ideology.

Less cynically, it is clear that Prof. Tabarrok's time and talent could have been better spent on more important topics. The suicide rate is really small, and whatever the causes, it cannot be substantially lowered without incurring other costs to society. I wonder if the housing bubble could have been mitigated if economists hadn't wasted their time on stochastically irrelevant but emotional distractions.

Boom, headshot!

No, that is not the point. The point is that we'd like to know about the causality between guns and homicide, but the direction of causality is difficult to ascertain. The causality between guns and suicide is much easier to ascertain. So the point is to establish causality in one domain, and then see how much we can apply that knowledge to a different but possibly similar domain.

Kudos to Alex for this study that will not please many who agree with him on other issues. It looks to have been well done and to have dealt with some of the more obvious possible criticisms. That said, lots of people just do not want to face up to this issue, which is often easy to do as homicides get lots of media coverage, whereas suicides rarely do, unless they involve famous people.

Regarding the cultural difference issue, this is a big deal. In Japan, where there is very low gun ownership (and a very low homicide rate), suicide is actually admired under appropriate circumstances, seppuku or hari-kari. A good samurai will disembowel himself if he brings shame, although tody people are more likely to hang themselves or jump off buildings. In any case, cross-country comparisons are essentially useless because of this cultural factor.

A large minority of the US population is Catholic, thus believing it is a sin, and most Protestant sects also strongly disapprove of it. Those suicide clauses in life insurance contracts are a strong disincentive for most people with families that do not hate them, and in many states it is actually illegal, weirdly enough.

While there are some alternatives as effective as using a gun, particularly jumping off tall buildings or the Golden Gate Bridge, not every locale has such opportunities. Most other options like overdosing on drugs, cutting one's wrists, engaging in a car accident, or sitting in one's garage with car on, have high failure rates, as high as 90% for some of them. Guns are very effective. I know quite a few people who attempted and failed to commit suicide who went on to live useful and reasonably happy lives, but if they had a gun would be dead from that momentary impulse.

Just to reinforce the data Alex is looking at, of the five states wtih the highest per capita gun ownership, four of them are also among the five with the highest overall suicide rates, and of the five with the lowest rates of per capita gun ownership, four of them are also among the five with the lowest overall suicide rates. This is a very robust relationship, also supported by many studies looking at households: roughly speaking, having a gun in the house raises the probability of somebody dying in that house by a full ordre of magnitude. The reasoning here is as obvious as the nose on your face, so if you are resisting it, you should think twice about why the quality of your thinking is so poor.

BTW, for those who think restrictions on guns are the end of civilization, I note that the only other nation that enshrines gun ownership as a constitutional right at the level the US does is Honduras. We compete with them for supplying guns to the Mexican drug cartels, whoop de doo, I am so proud of America on this! Of the 48 rights enumerated in the UN Declaration of Human Rights, gun ownership is not among them. It is just our thing and ours alone, oh, except for those wonderful Hondurans.

With regards to suicide clauses what fraction of the US population even has life insurance?

Most suicide clauses are along the lines of "if suicide is after two years" we will pay out. Beyond that, it seems patently ridiculous that someone calls off an attempt because they realized the insurance won't cover it.

It probably would only affect the cases where the person is committing suicide so his family will get the payment, or at least where that is one of the rationalizations.

Japan's suicide rate is almost twice that of the US.

Mexico has a constitutional clause protecting gun ownership.

What other facts did you get wrong in your screed above?

You cannot, in reality, by a gun in Mexico. The constitutional clause was made effectively void by statute many decades ago. There is exactly one gun store in Mexico. And you won't get the paperwork needed to buy a gun at that one store.

Serious question: do Mexican drug lords import their guns from the U.S.?

I meant to say of the household studies, dying by suicide probability rises an order of magnitude if there is a gun in the house versus there not being one.

BTW, the probability of dying by homicide if you have a gun in your house also apparently rises, although not by nearly as much as dying of a suicide. Family quarrels, accidents labeled as homicides (happened to a family member of mine), and criminsals entering houses and getting the gun and using it on the occupants are reasons for this.

The bottom line is that if people want to buy guns for hunting or marksmanship contests or just looking at them, fine. But anybody who buys a gun with the idea that they are protecting themselves or their family is seriously deluded and misinformed.

Just like nobody has 2.5 kids, nobody embodies statistical averages and they are wise not to act as if they do so. Probably for some people owning a gun is actually a good decision, and for others it is not. Some of those in the latter category probably think they are in the former category.

In general, when it comes to decisions affecting life or death (or home invasion rape/torture of your whole family) I'd rather let people make that decision for themselves rather than having it made by government bureaucrats.

Out of curiosity, at what point does the change in mortality due to not restricting this decision change that view?

It's an honest question. The answer clearly can't be "We need government intervention if there's any non-zero change in mortality," but it's got to be somewhere or everyone who wants to gets to build bombs.

So I guess what I'm saying is, the discussion has to be about where to draw the line, not if the line should be drawn in the first place.

"But anybody who buys a gun with the idea that they are protecting themselves or their family is seriously deluded and misinformed. "

LOL, you must be thinking of the long-discredited Kellerman study. Link:

Check out these seriously deluded and misinformed fools at these links:

Sorry, Kevin, but your links do not remotely discredit Kellerman, and there are other studies confirming Kellerman. These are anecdotes, and sure there are cases where people use guns to stop crimes, save themselves, and so on. But the issue is the total numbers, and your links say zero about them, and there are no studies I am aware of that seriously discredit Kellerman, although the findings have been modified.

It remains a hard fact, on average people with guns in their homes are more at risk of dying from gun homicides and much more likely to die from gun suicides, even if thi does not hold for a minority of those with guns in their households. The exact size of those effects can be debated, but that they are there really cannot be, and a bunch of anecdotes do not discredit it.

I have personally stopped three violent felonies over the last 29 years, because I had a gun available. Otherwise our life is quiet upper-middle/upper and peaceful. Stop putting armed criminals back on the streets of our nearby big city, and none of the three events would have occurred.

Penn produced a study in Philadelphia a few years back. It determined that people who carried a gun were more likely to be shot. But of course. A person under regular threat is much more likely to...carry a gun, as are people in the drug trade. Guns don't change the dynamic of lives and families, but are brought into the existing dynamics of lives and families.

What a colossal waste of time. Who cares about any of this beyond leftists partisans looking for a scintilla of evidence to stick it to conservatives with? Last I checked people own their bodies; they can off themselves in any way they see fit as far as I am concerned.

"Those suicide clauses in life insurance"

You mean the ones that say "wait a couple years before doing it"?

I have a rather different take. It will sound harsh, harsher than my personal emotions on the topic actually are.

To any study of the form "guns (fast food, cars, drugs, gambling) is/are associated with suicide (ill health, violence, low productivity)" ...

So What?

There is long and solid proof associating alcohol and various other drugs of abuse with all sorts of personal and social ills. And prohibition was abandoned, marijuana is now legal in some states, and in spite of spectacular enforcement efforts no credible observer thinks that drugs of abuse are unavailable in our society. We make it a royal pain to buy anti-histamine in order to thwart meth labs, but over even the medium term has this worked?

It's pretty easy to see that private cars cause a great deal of mayhem (though declining over time) - and there are in fact people (fools in my view) who would ban private cars. Clearly, cars are here to stay.

The issue with studies like this (on any of this whole class of topics) is that there is no useful actionable policy that will really affect society in a major way. But these arguments feed the parasite of political burden.... Politicians need some problem they can promise to *solve*, and are therefore sure every issue is a job for *them*.

You want fewer suicides, less drug abuse, and so on? Arrange for better preparation of people to be comfortable and successful members of our society. And be fully prepared that you will never totally succeed.

[And of course, Japan, a highly industrialized country in many ways comparable to the US, has very limited firearm possession and a sadly very high suicide rate.]

We just need some booze and pot insurance to offset particular losers in your decision to imbibe.

We definitely need booze insurance. If there were no booze on the streets Saturday night, there would be very few fights. Many fewer punches, clubbings, shootings, and knifings would occur. Cops know this, but the knowledge will lead nowhere. Indeed, this post is about guns and suicide. Tie in data on drinking and suicide, by gun or other, and you'll see a very strong correlation.

I hope when (if) I am old and facing a progressively debilitating illness, I am still permitted to own my own firearm with which to remove myself from this world on my own terms.

Firearms are used for 53.7% of suicides in the USA; suicide by jumping (despite all the window-ledge cartoons) is uncommon.

But in Hong Kong, jumping accounts for 52.1% of all suicides- perhaps because everyone living there has access to tall buildings.

Firearm suicides are more common in those parts of the USA where gun ownership and use are commonplace. Of course, firearms are usually easy to obtain in these areas as well. Even so, if one assumes that most suicides are impulsive then it does not follow that a firearm bought shortly before a suicide was necessarily bought for that purpose.

In short, I'd expect that easy access to any means of killing oneself that is relatively easy to commit, likely to result in death, and unlikely to produce prolonged pain or terror (e.g. fire, drowning) would not only correlate with suicide but might be a cause of it.

But if this is the case, then focusing solely on firearms does seem suspicious- why not consider the larger thesis instead?

I looked at gun suicide rates by state: one obvious factor is religion. My impression is that Catholics don't kill themselves as much as Protestants. Evangelical Protestants shoot themselves less often, all else being equal, than mainline Protestants, and unreligious men of Protestant background shoot themselves the most, ceteris paribus.

Also, whites kill themselves a lot more than blacks do.

What are the correlation coefficients?

A sizable fraction of gun suicides are by terminally ill men, but the newspapers very often leave out the immediate cause of death when somebody with a few months left to live can't take the tumor pains anymore.

Newspapers seldom mention suicides by the hopelessly ill, so the statistics on gun deaths can be startling, even to the worldly. For example, as part of a gun-control campaign back in 1989, TIME Magazine printed the stories of all 464 people killed by firearms in one week. I flipped through TIME’s portfolio on the outbound flight of a business trip I took with one of Chicago’s top lawyers. The attorney, an urbane fellow whose brother later became a US Senator, looked the magazine over on the way back, then articulated the same reaction I had left unsaid: “Wow, I didn’t know so many old guys shot themselves.”

Having provided some close care for two grandfathers and a stepfather, all afflicted with some form of dementia or Alheimers in their late eighties (and heart/circulatory ills), I am not surprised the well-informed choose a speedy exit. I wouldn't wish four or five years of diapers and confusion on any man after an otherwise decent and productive life. This is the down side of pneumonia vaccines.

Interesting, but ultimately pointless.

Suicide is a fundamental human right and any policy pursued toward the end of reducing it by force is absolute tyranny.

Cool that the Science Direct link also provides a deck of your presentation slides. I think it's excellent practice and suddenly I wish every journal would just require their authors to provide that upon acceptance of their manuscript. It's more informative than the abstract, and yet shouldn't cannibalize the publisher's business too much (it's only 7 slides).

Also: nice paper!

Will you share an ungated version for us in developing countries?

It would probably be good to mention the number of suicides and the percentage which are gun suicides.

Suicide rates from 1991 to 2009 on the CDC website. You can come to your own conclusions, but it seems to me that they may be rising after reaching a low in 1999, but the curve has been within a rather narrow range.

I've seen statistics showing gun ownership rates falling over the years, but I find that counter-intuitive since I think that we've all seen the news reports of purchases increasing. Maybe those are current owners stockpiling, which is true, but anecdotally there are many people I know who never owned a firearm before now making the decision to obtain one.

The incidence of firearms being used fell dramatically for the 10 - 24 age group to a low in 2009 along the same period and is creeping up slightly, while suffocation increased dramatically.

Trends in Suicide Rates* Among Females Ages 10-24 Years, by Mechanism, United States, 1991-2009

They don't seem to have the stats for males in the same age group. I can't say why.

Males tend to use firearms as the prevalent mechanism of choice, fell and rose along a shallow curve.

Trends in Suicide Rates* Among Males Ages 45-64 Years, by Mechanism, United States, 1991-2009

Women age 45 - 64 have a strong preference to poisoning, trend 2nd along a shallower curve than for males.

Trends in Suicide Rates* Among Females Ages 45–64 Years, by Mechanism, United States, 1991–2009

Leading causes of death:

How does these rankings figure into the matter:

Note how Cuba has a higher rate than the U.S., but I doubt private firearms ownership is particularly common there. France is ahead of us as well, and you can't just walk in to Cabela's and buy an AR there either. The number 1 country on that list is Greenland. I haven't checked, but I doubt gun crime is high on their list of social pathologies.

This leads to some questions:

1. Other comments have made the point of cultural factors in Japan/Korea/China on the acceptability of suicide. But if suicide patterns are culturally/religiously relative doesn't that mean they are inherently dynamic. If we could magically confiscate all firearms in the U.S. tomorrow i think Alex et al are absolutely right. Suicide rates would noticeably drop (and I'm a firm gun rights supporter). But how long would they stay down before equally "low-cost" suicide methods emerged to, ceteris paribus, bring them back up.

2. How constant are suicide rates over time, in this country and others; and how does that relate (or not) to guns. For instance, did suicide rates (with guns or other methods) go up or down after the 1934 NFA? Did urbanization, here or elsewhere, play a role, and so on.

3. How many people who kill themselves with a gun both a) bought the gun for that purpose and b) did not own a gun beforehand.

My point, both as a gun-rights supporter and as someone who tries to be thoughtful is, do we have enough information, especially historical information across cultures, to understand the problem, and what any particular solution (gun-related or not) would do to help it? I think that's at best an open question.

Something else for the gun control advocates to consider: a rusty, antique, single-shot, bolt-action hunting rifle is quite as effective for suicide as the most "tacti-cool" tricked out AR. Same dynamic for pistols and shotguns, and small calibres have nearly as much grisly effectiveness as large ones.

Given that there are between 270 and 315 million private firearms in the United States, how likely is it that the most common gun control measures proposed (magazine limitations, assault weapons bans, etc) are going to make a serious dent in American suicide numbers. If someone responds that we have to change the American gun culture over time to get the desired effect, and that these laws are a first start then I point you to my questions above. How long will that change take, and will no other "low-cost" methods emerge, or changes in America's overall suicide dynamic appear in the meantime?

Do you think that an increase in gun ownership increases suicide rates because it increases the effectiveness of suicide attempts? Men commit suicide at higher rates than women, even though women attempt at higher rates - I think because men use guns to a greater extent, whereas women swallow pills, etc. I volunteer on a suicide prevention hotline and have definitely seen this (the men mention guns much more than the women do).

Someone tell me if I'm not understanding the math right. These are rough numbers.

Gallup's best guess is ~40 % of Americans say they have a gun in the home:

The U.S. Census says there are ~117 million households in the U.S. as of 2010,

That's about 47 million households with guns.

So, given the suicide numbers Dr. Tarrabok quotes, there are .0008 suicides for every household with a gun. To get that additional .5 to .9 % increase in suicides 470,000 households that didn't have guns have to acquire them. Remember we are talking, approximately, between 200 and 350 additional suicides a year. Supposedly, the number of households with guns is going down: though I wonder how much of that is due to the fact that women are less likely to own a firearm, and there are a lot more women-headed households than 25 years ago, and not any drop in demand.

Given that the effect is real, and I tend to think it is, then it is small.

Again this leads to questions. How much could you increase gun ownership before the effect disappears, and numbers level off? Ten percent? Never?

Likewise, how much could you decrease them before the effect disappears?

And then there are intervening cultural societal changes that could drive the rate up or down (since clearly you don't need lots of privately owned firearms to have suicide rates a lot higher than the United States).

The more I think about this study the less I think it has to tell us, on either side of the issue.

Comments for this post are closed