What Should We Make of the New Data on Firearms?
By John Roman :: May 10th, 2013
Everyone agrees that the firearm debate would benefit from better data. In the last few weeks, several new data points have been released. Like much social science, the data show important correlations, but not necessarily causal connections. Thus, generalizing from these data is difficult. Here is what I think you would say about each, if you were trying to be scrupulously objective (which I am).
Let me begin with the most controversial.
Three days ago, the Bureau of Justice Statistics released a study showing that firearm homicides are way down, as are the number of non-fatal shootings. The report has not gotten much attention, only a handful of articles that are more partisan bickering then news. I’m surprised, because you don’t have to look at the graphic below for very long before a critical relationship becomes obvious; that is, that the period when firearms violence declines the fastest matches almost exactly to the period when the Federal Assault Weapons Ban was in place.
In 1994, when the ban was enacted, there were 17,527 firearm homicides in the United States. In 2004, when the ban expired, there were 11,624. In 2011, after seven years with no assault weapon ban, there were 11,101 firearm homicides, virtually unchanged from 2004. If you adjust for population growth, the change from 2004 to 2011 is slightly bigger: from 2.5 per 100,000 to 2.3 per 100,000.
As always, beware of the logical fallacy: post hoc ergo propter hoc (“before this, because of this”), which is just a fancy way of saying correlation does not equal causation. However, if you back the series out a little bit more in terms of years, you get an even more startling correlation. While we would like to look at firearms-related homicides before 1993, data on whether a homicide was caused by a firearm only go back to 1993. Therefore, we have to rely on homicide alone. But, since the decline in all homicides (my calculation) and the decline in firearms homicides were both 39 percent, it seems a fair assumption that the rate of homicide change approximates the rate of firearm homicide change.
When you extend the series back to 1960, the period where the assault weapon ban was in place is clearly the time period with the largest decline.
Of course, it must be noted that a lot of other things were going on: prison populations were growing rapidly, the crack epidemic and associated violence was declining, 20 years had passed since lead was removed from gasoline and abortion was legalized. And there are many other explanations. The problem is that many of these relationships are virtually impossible to unpack—for instance, prison population’s increase was caused by increased crime, so figuring out the effect of mass incarceration on crime is a tricky business at best.
And the evidence about the effect of the Federal Assault Weapons ban on gun crime is pretty weak (the evaluation of the Ban reported that the violence reductions were due to restrictions on the size of magazines). But it sure is a striking coincidence.
Unpacking Pew and polling data
On the same day as the BJS report, researchers at the Pew Charitable Trusts released some analysis of similar data, along with poll results about American's perceptions about whether the number of firearms-related crimes is growing or shrinking. The short version is that firearms crimes are decreasing, but the public thinks they are getting worse.
On April 8, the New York Times published an op-ed from two Democratic pollsters that reveals some startling facts about the gulf between gun law perceptions and reality. They summarize it quite succinctly: “Americans don’t really know what gun laws are on the books and [we] falsely construe that to mean they don’t want common-sense gun laws passed — when they clearly do.”
The findings of that poll are this: By a slight majority, Americans favor better enforcement of existing laws over new gun control laws. However, among those who favor better enforcement, about half believe background checks are currently required to buy a gun at a gun show or in a private transaction (which is only true in a handful of states).
A majority of those polled believe a gun cannot be sold to someone on a terrorist watch list (it can). One-third believe law enforcement is notified when large numbers of guns are purchased in a short time (it is not). Almost half believe ammunition cannot be legally bought over the internet (it can).
The problem, however, is that the pollsters are partisan and many wonder what that means for their results. There is no analogous poll by a Republican pollster, and the nonpartisan Pew poll does not ask these questions.
Finally, deep in the Pew data, there is support for a hypothesis first put forth by noted criminologist Alfred Blumstein: that young African-American youth with cheap handguns were responsible for much of the spike in violence in the 1980’s. Indeed, the largest declines in firearms-related homicides are among 12 to 17-year-olds, and declines among African Americans are larger than any other group. But without having the data, it would be an ecological fallacy to generalize to individuals from these general trends.
So what then are we to make of all these new data? My opinions—and they are only that—are that:
- The Federal Assault Weapons Ban’s contribution to the crime decline was real, but modest (but it could be made bigger).
- The opinion poll is probably right that Americans believe all kinds of laws are on the books that aren’t.
- Blumstein’s hypothesis is probably correct.
But until we can get better data on the correlation between violent crime and the availability of fire arms, we’ll all be relying too heavily on opinion and partisan rancor.