There has been much ado, among people who make much ado over such things, about a new study in the American Journal of Medicine that debunks a favorite claim of the gun lobby, the deterrence hypothesis — that is, the idea that high gun ownership drives down crime rates.
The deterrence hypothesis (whose adherents would be offended by my use of “hypothesis”) divides the world into two groups: “law-abiding citizens” who go about their daily lives without being violent, and “criminals,” predatory creatures who survive by victimizing the law-abiding. By arming the law-abiding, we scare the criminals straight or at least convince them to move to horrible places like New York where the helpless citizenry is disarmed. And to prove it, proponents of this idea point to American data on crime rates and gun ownership, which tend to show that where gun ownership is higher, crime rates are lower.
There are a number of problems with this idea, among them the fact that gun ownership is higher in rural America, where crime is lower regardless of the supposed deterrent effect, and that its dichotomy of good guys and bad guys is the property of an eight-year-old’s mind. But no matter; in the great culture-war tradition of refusing to talk about the real issue — differing values and worldviews — the other side is keen to find proof that the deterrence hypothesis is flat-out wrong.
Enter our new study. In summary, the authors took data on gun ownership, crime rates, and mental health in 27 developed nations and fed them through the magical statistical machine of regression, which led them to conclude that gun ownership is correlated with higher firearms-related deaths to a greater degree than is mental health. Correlation is not equal to causation, they sagely advise, but it sure looks like deterrence is full of shit, as is the claim that mental health is the real problem. And that’s true — but the people making much ado about this study then go on to claim that it demonstrates that higher gun ownership leads to higher crime, a claim that is probably full of shit.
The fact of the matter is that this study is neither truly new, nor very good. Previous research has used homicide rates and gun ownership from similar sources to demonstrate that homicide rates are higher in countries where gun ownership is higher. Gary Kleck has criticized that work, noting that when we discard the single outlier — the United States, with its outrageously high gun ownership rates — the supposed correlation disappears. Much as I am inclined to take Kleck with a grain of salt, his point here stands, and it leads me to sniff this new study carefully, and to note an odour of fish.
The problem with any such analysis is in the sample, of course, and two things about the sample leap out. The first is the choice to restrict the analysis to 27 developed nations. While that’s defensible, because we don’t want to confuse the matter by mixing stable democracies with countries racked by political violence, it also leads us to discard all of the potentially confounding evidence. We know that the US has high gun ownership and a high rate of violence, and that Japan has low gun ownership and low violence, and that the other guys are somewhere in between, and we can be pretty confident of the results simply by selecting that sample and ignoring (for example) Jamaica, where private gun ownership is banned but the homicide rate makes Detroit look peaceful. One might argue, of course, that we exclude those countries because we suspect their rates of violence have other causes, but this in turn suggests that we ought to be controlling for a lot of “other causes” — which we are not.
The second is the importance of two obvious outliers, the United States and Japan. The US has the highest rates of gun ownership and mental illness, and Japan the lowest rate of gun ownership and a low rate of mental illness. The US also has the highest rate of firearm-related deaths, and Japan the lowest — a disparity so marked that we have to use a logarithmic scale to plot the regressions. So here’s a legitimate, obvious question: what happens to the strength of this correlation when those two outliers are removed?
We might also ask whether the rates of mental illness, which come from WHO data that relies on local diagnosis rates, are truly reliable. Is the disparity between the American and Japanese rates truly reflective of their national mental health, or might it also reflect differing cultural attitudes towards mental health? And can we really say that this proves mental health is not the problem?
Probably not. And this is the problem with much of the “proof” bandied about in the Great American Gun Kerfuffle: it really isn’t proof at all. It is the same old data, the same old ideas, and the same old arguments, endlessly lobbed back and forth to no avail. No one is changing his mind, because we are failing to address the truth of the matter. Gun ownership in the United States is a cultural phenomenon. The gun is a symbol that stands for, or against, a vision of how Americans ought to live and what kind of a place the United States ought to be. No one is willing to talk about this frankly, because that’s a conversation about values. And nobody wants to admit that the problem of American violence is primarily a problem of values — of America’s fondness for the idea of violent self-determination. No one wants to consider that the gun is a symptom of a deeper problem, not a cause. So America retreats into its war of statistics and data, and gets nowhere.