« An interesting survey | Main | Only government actors are safe with arms, part 364 »

CT study that supposed showed 1995 law reduced homicides

The study, as usual, got play in the press... and then the bloggers took it apart.

Here is Robert VerBruggen's take. Homicides and gun homicides were plummeting nationwide, beginning in 1994, and presently are about half what they were at that peak. So how to measure whether a 1995 law worked? How to determine whether homicide rates would have fallen more, or less, if the law hadn't been enacted?

The creators of the "study" compared the real CT to a manner of artificial CT assembled out of portions of the neighboring States of Rhode Island and Massachusetts, and concluded that the rates in the real CT fell faster and more deeply than the rates of those areas. But, VerBruggen points out, RI had an atypical murder rate surge during the period of the study (it was normally below the national rate, but shot up to well above it; MA's rates generally rose proportionate to the national, as well), which operated to inflate the "artificial CT" homicide rate. Moreover, compared to national rates, CT's rates, compared to national rates, surged to a high level just before the study began, and thereafter declined to their usual relationship (before beginning another surge, just after the period the study looked at. (In addition to all else, this suggests that the study was cherry-picking its data, both in terms of timeframe -- ten years, when we have have twenty -- and in terms of creating the "artificial CT").

And here is John Lott's disassembly of the "study." He notes that the CT homicide rate began falling, and fell more steeply, in the two years before the law, and that both national and regional homicide rates fell more steeply than that of CT. (The latter illustrates the study's "cherry-picking." Comparing the rates to national rates or regional rates would have been much easier than creating an artificial CT, but wouldn't have yielded the desired result).

There's a reason that this, and all the other questionable studies, have been sent to medical journals rather than ones devoted to criminology. That way, the peer reviewers and editors are mostly MDs, familiar with disease but not with crime trends, data, and past literature. I'd be surprised if any reviewer or editor knew, for instance, the simple fact that murder rates have been falling nationwide for the past twenty years.