In discussing gun control in “Gun Science” [Skeptic], Michael Shermer first cites a 1998 paper in the Journal of Trauma and Acute Care Surgery concluding that guns in the home are much more likely to be used in criminal assaults or homicides than for self-defense. Oddly, that study accounted only for cases where criminals were killed or wounded and not the more typical scenario in which an attacker is scared away. Cases where the attacker is killed or wounded account for well less than 1 percent.
Shermer ignores the 2004 National Academy of Sciences report that rejected the 1998 study and similar ones. As the NAS report noted, this type of public health research fails to account for the endogeneity problem—that it is especially people who feel threatened who tend to acquire guns. Fixing this reverses the claims.
John R. Lott, Jr.
Author of More Guns, Less Crime, third edition (University of Chicago Press, 2010)
SHERMER REPLIES: The 1998 paper is one of several that determined how often guns in the home are used to injure or kill in self-defense, compared with how often they are involved in accidental shootings, criminal assaults and suicide attempts. Tragic shootings outnumber defensive ones by more than 40 to one. In addition to cases in which a homeowner chased a bad guy away, the 1998 study did not count how often guns in the home were used to threaten or to intimidate a family member, a spouse, a girlfriend or a neighbor.
The self-defense figures Lott cites were derived by extrapolating low-frequency responses to public opinion polls to the entire U.S. population and are thus wildly inflated. Police reports suggest the use of guns in self-defense is much less common. An audit of Atlanta Police Department reports of 198 home-invasion crimes identified three cases in which victims successfully defended themselves. Intruders got to the homeowner's gun twice as often.
As for the NAS report: when the committee opined about case-control research, it was criticizing an analytical method most, if not all, of its members had never employed. The endogeneity issue is a case in point. The committee speculated that any statistical association between guns in the home and violent death may exist because people acquire firearms in response to specific or perceived threats or because gun owners may be more or less violence-prone.
But Arthur Kellermann, the lead author of the 1998 study that had been referred to in the committee's report, has militated against that possibility*. For instance, in a 1993 case-control study, he questioned households about a wide range of risk factors for violence and took any differences into account through logistic regression. In addition, every household was matched with a control in the same neighborhood, which ensured similar socioeconomic status and exposure to crime.
Interestingly, the study found that a household's risk of homicide from an intruder was neither higher nor lower if a gun was kept there but that the risk of homicide from a family member or an intimate acquaintance was much higher.
In discussing potential meteor strikes in “Preventing the Next Chelyabinsk” [Advances], John Matson writes, “Fortunately, impacts on the scale of Chelyabinsk occur only once a century, so perhaps humankind will have figured out even better techniques by then,” thus falsely supporting the common misunderstanding that if something happens once a century, it won't happen again for 100 years. That would require collective memory among meteors.