Are gun suicides gun deaths?

After the President asked news organizations to compare deaths by firearms with deaths from terrorism, many did. It’s pretty striking, here’s the chart from CNN:



But not all of the charts looked the same. Vox’s chart excluded suicides, and thus came to a much smaller (though still striking) result:

gun homicides v terrorism deaths

And this raises an important question. Should we include suicides with a gun in our calculations of gun deaths? Asked that way, the answer seems obvious: of course they should count, a gun death is a gun death. But I think there are important reasons to exclude them. Suicides make up 2/3 of all firearms deaths, but those deaths are probably not preventable by restricting access to guns.

Here’s the argument in outline: people commit suicide with guns because that’s the dominant image of suicide in our culture. Serious restrictions on gun control would change the dominant image of suicide in our culture, and potential suicides would choose different methods.

And yet there’s plenty of evidence to the contrary, so it’s important to think that evidence through. In many cultures, targeting the means of suicide led to major reductions in suicide. We know that when we interview survivors, we find that 70% spend less than an hour planning, so they’re mostly using ready-to-hand methods. Britain reduced its suicide rate by changing the kinds of gas they used to power their ovens. San Francisco hopes to prevent suicides at the Golden Gate Bridge with a steel net.

And yet there are many differences between survivors and successful suicides. Successful suicides plan more carefully, and for longer, choosing deadlier methods. Successful suicides are generally older and maler than survivors. In short, successful suicides seem to mean it more than unsuccessful ones, they make determined choices based on social facts rather than psychological whims. So they won’t be deterred if forced to change methods; we need to change the social facts instead!

The National Academies of Science concluded much the same thing when last they evaluated these questions: “Some gun control policies may reduce the number of gun suicides, but they have not yet been shown to reduce the overall risk of suicide in any population.” The difference is between micro-reductions (fewer deaths at the Golden Gate bridge) and macro-reductions (fewer deaths per 100,000 people.) We can make lots of changes at the margin without touching the infra-marginal causes of suicide, which are likely not tied to method.

This is an example of Simpson’s Paradox: you can reduce the marginal propensity to commit suicide, even make small, local reductions in those numbers, but without affecting the overall rate of suicides. And the same thing goes in the opposite direction: increasing access to guns might lead to short-term and local effects on suicide, but we compensate in various ways. Otherwise, it would be hard to make sense of the fact that our suicide rate is quite low while we have more guns than anyone in the world. We’re in the middle of the pack for the OECD, and France, Finland, and Japan all beat us:

suicide rates

Why is the US suicide rate so low, given the overwhelming number of guns we have? Guns seem to massively increase our homicide rate but not our suicide rate, so what’s going on there? For the largest group of suicides: people basically understand the lethality of the method they choose, and they use less-lethal methods if they don’t want to succeed. So suicide success mostly tracks desire, not access.

Around the world suicides use different methods:

“Poisoning by pesticide was common in many Asian countries and in Latin America; poisoning by drugs was common in both Nordic countries and the United Kingdom. Hanging was the preferred method of suicide in eastern Europe, as was firearm suicide in the United States and jumping from a high place in cities and urban societies such as Hong Kong Special Administrative Region, China.”

So while of course, there’s some feasibility to reducing access to pesticides and firearms, there’s little hope of eliminating access to drugs, high places, or potential nooses.  Britain removed a major non-violent method, and saw serious reductions. We can do the same by targeting non-violent suicide methods. But the big group of suicides have a high transition rate among violent methods (hanging, firearms, jumping) while the smaller group of suicides have low transition rates among non-violent methods (ovens, drug/pesticide poisoning.)

So you can best reduce suicide by regulating access to pesticides and drugs, but regulating access to firearms leads to substitute violent methods. There’s even the risk that though relatively few people will transition from non-violent to violent methods, they’ll do so at a much higher success rate. Britain avoided that fate with its coal gas oven transition, but further targeting prescription drug poisoning might yield different results.

There’s nothing wrong with trying to make suicide more difficult; I endorse strategies that might force rethinking, like waiting periods. But I’m not convinced it’s possible to deter the determined. So in that sense, we should probably treat gun suicides more as suicides and less as gun deaths. The best evidence is that policies designed to reduce our (already low) suicide rate will target social facts, about masculinity, bonding social capital, mental health and self-worth, while the policies best tailored to reduce our (quite high) homicide rates should target a single variable: access to firearms.





Second Opinions