Places With More Guns Don't Have More Homicides
In a new Voxsplainer on gun violence, Dylan Matthews claims:
Protestations of gun rights supporters aside, public health researchers who study firearms generally agree that increased firearm ownership rates are associated with higher rates of homicide. ... Developed countries with more guns generally have more homicide; states within the US with more guns have more homicide ...
The two assertions at the end are not true, and the first sentence explains why: If you want to give a good account of a debate about gun statistics, you don't treat the consensus of "public health researchers" as gospel. The field is notorious for its anti-gun bias, and there's a whole literature of work outside of it.
It's true (as Matthews notes) that there are some studies showing guns to be associated with increased homicide once other factors have been statistically "controlled" (a highly subjective process that can be manipulated, even subconsciously, to make the data say whatever the researcher wants them to say). But there is no simple relationship between gun ownership and homicide rates as such, either among developed countries or among states in the U.S.
I did the math on developed countries last year. Looking at a collection of countries that had been featured in a poorly conducted anti-gun study, I found that there were two major outliers: South Africa (which has a pretty typical gun-ownership rate but an astronomically high homicide rate) and the U.S. (which, as Matthews notes in a different "card," is high in both gun ownership and homicide -- though Matthews does not mention the biggest explanation for the high homicide rate, which is that the U.S., like South Africa, has a large, historically oppressed black population with high crime rates). When those two countries are booted from the analysis, this is what we end up with:
There's no statistically significant relationship, though I'm sure you can get one if you mess around with which countries are included (especially the U.S., which by itself can swing the data quite a bit, because that's what outliers do).
What about states? I took these 2001 gun-ownership numbers from the Washington Post (based on a risk survey) and crossed them with 2001 murder and nonnegligent manslaughter rates (per 100,000) from the FBI's Uniform Crime Report. (Spreadsheet here.) The result:
Again, if you want to go beyond asking whether "places with more guns have more homicides" (as Matthews's card is titled), you can control for other factors as well, or try using other methods. (For example, "public health researchers" seem awfully fond of "case-control" studies, which don't even work that well in epidemiology, the field they come from.) Then you can make a case that guns increase homicide -- but it's not that difficult to make the opposite case instead, if that's what you want to do. This just tends not to be the case for "public health researchers."
Robert VerBruggen is editor of RealClearPolicy.com. Twitter: @RAVerBruggen