The Never-Ending War over Gun Statistics :: 08/09/2020

The debate over gun control is in part a debate about fundamental rights, the things that a government should never deny to the citizens to whom it is supposed to be subordinate. Plenty of us conservatives would never surrender our right to keep and bear arms on that principle alone.

But it’s a heck of a lot easier to win this debate if gun control doesn’t reduce crime or is even counterproductive — and as a result, much of the argument takes place in the realm of academic studies and arcane fights over how to analyze data. For decades upon decades, the two sides have fought each other vigorously, creating lots of statistics for their followers to cite but resolving very few issues in a definitive way.

Those looking for a handbook of pro-gun stats should check out Gun Control Myths, the latest from John R. Lott Jr. The author is a veteran of the gun-research wars: A quarter of a century ago, Lott’s famous “more guns, less crime” study set off an enormous wave of research into concealed-carry laws — which, he found (and others naturally disputed), reduced crime by allowing the law-abiding to arm themselves against criminals. Now he heads the Crime Prevention Research Center, which has continued to monitor these laws and issues studies on other crime-related topics as well.

The book weighs in at fewer than 200 pages, including extensive charts and footnotes, and is hardly a balanced review of the gun-control literature. I’m sure a gun-control supporter could write a book twice as long taking issue, in painstaking detail, with everything it says. But it nicely pokes holes in the common narrative in which gun control obviously works and guns are the only reason America has a high murder rate. Above all, it shows that there’s another side to just about any factoid you encounter in the mainstream media.

There’s page after page of statistics here, but to start, let’s look at one very simple question: Do countries with more guns have more violence? (Forget about whether guns cause violence, in other words; we’re just asking whether the two go together.) This is the sort of thing any high-schooler could put together in an Excel spreadsheet — but there are lots of different ways to do it.

Here, for example, is a chart you’ve likely seen on social media at some point. It tells a very tidy and alarming story: Places with more guns have more gun deaths, and the U.S. has boatloads of both.

But there are a number of easily disputed decisions behind this chart.

For one thing, it focuses only on deaths by gunfire. It’s almost tautological that places with more guns will have more gun deaths — after all, a place with zero guns can’t have any gun deaths — but that doesn’t mean these places have more violence in total. Without guns, people might switch to other weapons instead, and the physically weak might be unable to defend themselves against stronger attackers. When we look only at “gun deaths,” we ignore these possibilities.

Second, “gun deaths” include suicides and accidents, not just homicides. Indeed, most of the deaths are suicides, and the distinction is highly relevant to any debate over gun policy. Taking away someone’s freedom on the off chance they might hurt themselves is very different from restricting guns so people can’t harm others.

And third, there’s the question of which countries are included. The chart covers only a select group of developed countries, raising the question of which countries are developed and which are not. In addition, some countries one might include, including the U.S. itself, are clear statistical “outliers” that would normally be trimmed from the data.

Make a few tweaks to this methodology and the entire relationship dissipates, as Lott shows:

It’s certainly true that the U.S. has a lot of guns and a lot of homicides. If we added it to this chart (as in another graph Lott includes), it would fall somewhere around Chile in terms of homicides, and it would force us to extend the gun-ownership axis up to about 90. But if gun ownership increases homicide in general, the effect simply isn’t strong enough to show up on a chart like this. Other factors overwhelmingly drive the immense differences in homicide rates we see across the world. Digging further into this question requires more sophisticated research, which inevitably requires even more subjective decisions on the part of the researchers.

If you can make data like these — death rates from government sources — look however you want, it’s even easier to change the results when you’re counting the deaths yourself. In a heavily publicized study, the researcher Adam Lankford purported to count all the mass shootings across the world since 1966, and he alleged that the U.S. had an incredibly disproportionate share of them: We had 5 percent of the world’s population but nearly a third of the mass shootings. He promoted the results to sympathetic media outlets before the paper was published but refused to share the data with critical researchers until much later.

In response, Lott put together a competing data set covering only the period since 1998 (for which records are much closer to comprehensive), published it openly from the get-go, and found something much different. By his tally, the U.S. does not have a high rate of mass shootings, and the worst-hit countries are places such as Iraq, Afghanistan, and the Central African Republic.

And when the murky reality of gun violence is filtered through the media, everything gets a little more distorted yet. In one chapter, Lott chronicles case after case in which a civilian carrying a gun interrupted a mass shooting. Few of these received much national media attention, despite the clear newsworthiness of heroes putting themselves at risk to save others’ lives.

These, of course, are the simple things: how basic charts are made, how we count violent incidents, how the media chooses to cover shootings of different kinds. But most of the studies you read about these days are not so simple. They involve using complicated methods to see how crime rates changed when gun laws changed, taking into account how crime changed in places that didn’t change their laws — and for good measure, “controlling” for other trends that might affect the results. There are any number of ways to put a study like that together, and all of them will lead to different results.

I’ve been following the statistical back-and-forth over gun control for about a decade and a half, and for me the upshot is that if gun laws have an effect on violence one way or the other, that effect must be small enough to be highly sensitive to how you measure it. It’s not as if these laws, up to and including aggressive gun bans we could never implement here, instantly and undeniably cause crime to skyrocket or fall. With that in mind, and given my commitment to gun rights as a matter of principle, I’m open to measures that aim to keep guns specifically from criminals, but highly opposed to any effort to disarm law-abiding citizens.

The moms demanding action and the guys wearing “GUNS SAVE LIVES” buttons obviously have much stronger opinions about the empirical questions here. I suspect they will be hurling numbers back and forth at each other for decades to come, because Lott and his fellow gun researchers are not going to resolve their many differences any time soon.

https://www.nationalreview.com/2020/08/book-review-gun-control-myths-handbook-useful-numbers/