New Study: More Guns, More Homicide
You can see the abstract here, and it's been posted online in full here. The researchers looked at state-level data going back to the 1980s, and found that gun ownership (measured through a proxy, the percentage of suicides committed with a gun) is positively correlated with firearm homicides after controlling for various complicating factors. In English: More guns, more gun homicides.
Upon seeing the results, my first reaction was that it doesn't make sense to focus on firearm homicides. For example, when gun ownership drops, people who lack access to guns might kill with other tools instead -- this produces a drop in firearm homicides but doesn't actually save any lives. However, when I e-mailed Michael Siegel, the study's lead author, he provided this explanation:
We chose to examine firearm homicides because the hypothesis we were testing is that the availability of guns would specifically cause an increase in firearm homicides. However, I did run the model looking at the overall homicide rate as an outcome variable, and it does turn out that the gun percentage proxy is positively and significantly associated with total homicides as well. The relationship between gun ownership proxy and non-firearm homicides is positive and significant, but much smaller than that with firearm homicides. But the bottom line answer to your question is yes, we did find a significant association between gun ownership proxy and total homicide rates.
In other words, in this model, states with low gun ownership have significantly lower rates of gun homicides and don't have higher rates of non-gun homicides to make up for it; in fact, they have slightly lower rates of non-gun homicides, too.
There are some big words of caution, though, with any study that uses statistical techniques to "control" away confounding variables. The paper itself concedes it does not establish causation; it's possible that some variables are left unaccounted for, or that gun ownership is a response to homicide rates rather than a cause of them. (Regarding the latter, however, the results did survive when the researchers "lagged" the data by up to two years.)
And on a more cynical note, I've been covering the gun debate for more than half a decade now, and I've learned not to be overly impressed with this kind of work, however much I respect the technical chops involved. When this many variables need to be sorted through and this many methodological decisions need to be made, different researchers will reach different results. If you're a gun-rights fan looking for a study to cite in response to this one, just wait.
Robert VerBruggen is editor of RealClearPolicy. Twitter: @RAVerBruggen