Do Guns Cause Violence?

At The Week, my old friend Michael Brendan Dougherty makes a "conservative case for reforming America's sick gun culture." (He and I were roommates in a disturbingly messy four-bachelor Fairfax townhouse about a decade ago.) He supports the idea of an armed citizenry, and believes people should be able to own guns to protect themselves. But in light of yesterday's events, he would make it a requirement for anyone buying a gun to have some sort of training or socialization in the culture of gun clubs.

Of course, this is a political nonstarter — it amounts to background checks on steroids, where not only your criminal history but also your character and training can disqualify you. We do have criminal checks for gun-dealer sales (and yesterday's killer apparently passed one), but attempts to require these checks on sales between private individuals, a modest measure I tentatively support, have flopped outside a handful of liberal states. It's hard to see Dougherty's plan faring even that well. And then there's the question of whether such a rule would square with the Second Amendment as the Supreme Court interprets it.

But it's worth engaging with Dougherty's assumptions about the broad link between guns and violence, because they're shared by vast swathes of the population. So, here's some of the wisdom I've accumulated in more than a decade of following gun research.

Dougherty writes:

America is really the only nation that is orderly with an almost unchallengeable state, and yet has a gun-death rate similar to much poorer Latin American nations experiencing low-grade civil wars and disorder.

Yes, many of our firearm-related deaths are suicides. But our firearm-related homicide rate is noticeably higher than every comparable industrialized nation. And furthermore, there seems to be a strong correlation between reduced access to firearms and a reduced rate of suicide.

"Gun deaths" are a pet peeve of mine, and Dougherty only partly addresses my concern when he admits that they include suicides. The notion that guns and "gun deaths" go together is practically tautological, and unhelpful to boot. A country with no guns by definition has no gun deaths, but that doesn't mean it has fewer violent deaths overall.

To start with a point of agreement, I'm somewhat sympathetic to his point about suicides. Unlike with homicides — where a gun can enable one or prevent one — the effect of guns on suicide can only be bad. There's decent research suggesting that gun ownership does modestly increase suicide rates; suicide can be impulsive, so it's not true that someone without a gun will necessarily find another way. But many people — including me, and I'm guessing most conservatives in general — find repugnant the idea of reducing people's "access to firearms," not on the basis of any demonstrated suicide risk, but simply on the off chance that they might use a gun to harm themselves.

If not "gun deaths," what about "firearm-related homicide"? This too is a nearly useless concept, because gun homicides and non-gun homicides interact with each other. Someone who can't get a gun may simply kill with a different weapon instead. (Even in gun-drenched America, about a third of murders are committed with no gun.) And someone who can get a gun might defend himself against an assailant who doesn't have one. So we should always focus first on total violence, not gun violence, even when we're looking for the effects of guns.

The simple correlation between gun ownership and violence often disappears entirely when you take this into account, as I've shown with data on both states in the U.S. and developed countries. This shows that guns are not a primary driver of differences in murder rates — whatever effect they have is drowned out in the data by things like demographic differences, culture, and so forth.

Using complicated statistical techniques, you can try to tease the effect of guns out of this mess, and some researchers have purported to do so. But as statistical techniques become more complicated, they also become more subjective and run the risk of falling victim to political motivations. The two fundamental laws of gun studies are: One, if a given author reaches a pro- or anti-gun result in one study, all his future results will point in the same direction; two, if it appears in a public-health journal, the results will suggest guns are bad. Relatedly, a general note of caution is always in order when it comes to social science: It's impossible to "control" for everything besides guns that might affect violence, especially culture.

Essentially, the tools currently available to scientists aren't precise enough to resolve this debate, leaving too much wiggle room for researchers to reach the conclusions they want. We don't have consensus, but rather groups of researchers reaching conflicting results. Here's a criticism of the study linked above, for example. 

We see a similar thing in the debate over shall-issue concealed-carry laws, under which any civilian without a serious criminal record can get licensed to carry a gun. Some state laws are incredibly permissive — a few don't even require permits or training, and I got my Virginia license on the basis of a Wisconsin hunter's-safety certification I earned when I was 12. For all the state knew, I hadn't touched a gun in more than 15 years.

This would seem to be a prime example of the anyone-can-pack-heat culture Dougherty wants to reform. But as with the research on gun ownership, 20 years of studies on these laws have taught us almost nothing. Some studies suggest the laws reduce crime. Others suggest they have no effect. Still others say they increase crime. And even the most recent study reaching the anti-gun conclusion admitted that the results are incredibly sensitive. The most the authors could say is that the results are anti-gun if you use the techniques they happen to prefer.

I said we've learned almost nothing. What we have learned is this: A bunch of states started letting almost any random person walk around with a gun, and if anything good or bad resulted, it doesn't reliably show up in the data. That's something in itself.

Other ways of studying gun restrictions are even less conclusive. For example, the "public health" crowd is quite fascinated by "case-control" studies, where they compare people who got murdered with demographically similar people who didn't get murdered, and pretend it means something that the people who got murdered were more likely to own guns. And studies looking at states before and after they implemented gun-control measures range from interesting if only suggestive to laughably bad.

I'm not the only person to reach the conclusion that the role of guns in violence is rather subtle. One interesting example is the Harvard psychology professor Steven Pinker. He's no fan of the NRA; he's from Canada, for God's sake. But in his book about the decline of violence, The Better Angels of Our Nature, the discussion of "weaponry and disarmament" is practically a footnote — about one page in an 800-page tome, relegated to a section about the "forces that one might have thought would be important [in major trends in violence] ... but as best as I can tell turned not out to be." He doesn't even bother to "endorse the arguments for or against gun control," and he writes that "human behavior is goal-directed, not stimulus-driven," adding that "anyone who is equipped to hunt, harvest crops, chop firewood, or prepare salad has the means to damage a lot of human flesh." Similarly, in Ghettoside, her interesting exploration of black-on-black crime in LA, the journalist Jill Leovy writes — in an actual footnote — that "guns are not a root cause of black homicide." The criminologist Gary Kleck tends to be highly skeptical of claims that guns make a difference, on net, one way or the other.

In short, yes, it's possible that confining gun ownership to the people willing to jump through various government hoops might have some marginal effect on violence. But that effect will probably be so small as to be difficult to detect, and there may be no effect at all.

Robert VerBruggen is editor of RealClearPolicy. Twitter: @RAVerBruggen