March 01, 2002
From the desk of Jane Galt:Via James Rummel: a Harvard
Via James Rummel: a Harvard researcher has come out with a study that says that children die at higher rates in states with higher gun ownership. Certainly provocative, although as Rummel points out, this same researcher has given us such research gems as "smoking causes suicide" (the good doctor does admit, grudgingly, that the causal relationship might be reversed), and "binge drinking leads to gun ownership", so as Rummel says, the one piece of rock solid data we can glean from this study is that its author, Dr. Matthew Miller, doesn't like drinking, smoking, or guns.
Still, some of the evidence is facially convincing:
The study showed that the five states with the highest gun ownership levels had many more firearm-related deaths among children than the five states with the lowest levels of gun ownership.The two groups of states had almost the same number of children, but in the high gun-ownership states there were 253 accidental firearm deaths compared to just 15 in the low gun-ownership states.
There were 153 firearm suicides in the high gun-ownership states compared to 22 in the low-ownership states and there were 298 firearm murders in the high gun-ownership states compared to 86 in the low-ownership states.
Meanwhile, the rates of non firearm-related suicides and murders in the two groups of states were much closer, leading Miller to conclude the increase in deaths was attributable to the higher number of firearm-related deaths.
Of course, any gun statistics for either side have to be taken with a shaker of salt, because the reporting problems are so difficult. For one thing, whether an incident gets reported as a murder, a suicide, or an accident varies tremendously from precinct to precinct; the best guess is that many gun "accidents" are actually homicides or suicides. So there's the possibility, for instance, that states with higher levels of gun ownership are less likely to give credence to the idea that "it just went off by accident" then are states with lower levels. I note that the article mentions only the levels of gun-related suicides or homicides; from the fact that it leaves out accidents, I infer that these are in fact higher (since similar or lower levels would have bolstered the argument). This seems to bolster the possibility of reporting error.
I also note a little oddity lower down in the article:
The difference remains even when the data is controlled for poverty, education and urbanization, the study found."Although no conclusions about cause and effect can be made, this study provides compelling evidence that states with high firearm availability are states with high childhood firearm death rates," Dr. Therese Richmond of the University of Pennsylvania's Firearm Injury Center wrote in an editorial.
The five states with the highest rates of gun ownership are Louisiana, Alabama, Mississippi, Arkansas and West Virginia. The five with the lowest are Hawaii, Massachusetts, Rhode Island, New Jersey and Delaware.
How on earth would you control for urbanization between five almost entirely rural states, and five states of which four are on the Northeast Corridor and one is a resort mecca? There is a point at which controlling your data reduces your sample size too far to produce good results; think of the study which showed that gun ownership caused suicide, in which "controlling" for location discarded 2/3 of the data points.
Which is not to say the study is wrong. Indeed, at the margin I would assume that higher rates of gun ownership increase accidents, just as does higher rates of car ownership. The question, of course, is whether or not the benefits of gun ownership offset the risks; John Lott and others say they do, but of course their data has the same kind of reporting problems. I will be interested to see what the inevitable critics make of the data.
But I'm not going to listen to anyone who wants to argue that guns change marginal behavior in owners, but not in criminals.
Posted by Jane Galt at March 1, 2002 09:45 AM | TrackBack | Technorati inbound links