In 1993, Dr. Arthur Kellermann and several coauthors released a study in the New England Journal of Medicine comparing the ownership of firearms and likelihood of a homicide occurring within the home. In his findings, Kellermann reported that firearms ownership carried a 2.7 times greater risk of homicide in comparison to homes where firearms were not present. Since this is probably the closest thing to a legitimate study that gun control groups have, its findings are often recited like gospel anytime the issue is brought up in debate. Unfortunately for these groups, the study has some serious flaws that jeopardize its legitimacy and which have been well exposed by other researchers, such as John Lott, Ph.D.
In Chapter 2 of his book More Guns, Less Crime, Lott dedicates several paragraphs to a thorough critique of Kellermann’s work. The first and most significant of these criticisms is that Kellermann’s study fails to report that in only 8 of the 444 examined homicides could it be determined that the gun involved in the crime had been kept in the home and not brought in from outside. Part of the reason for this is that Kellermann only asked relatives of homicide victims whether or not a gun was kept in the home. He did not take this a step further to see if the gun that had been kept in the home was actually used in the commission of the homicide. Kellermann found some correlation between homicides and gun ownership, but never investigated whether or not the victims were shot with the gun held in the home, nor did he sufficiently establish whether the individual legally possessed the firearm or if that person had been involved in any other high risk/illegal activities. While he did ask relatives and neighbors of victims about potential drug or alcohol problems in the household, he did not thoroughly investigate other contributing factors that could have led to the crime.
In the same chapter, Lott compares the Kellermann conclusion to a hypothetical study assessing mortality rates between those who had recently visited a hospital and those who had not. Lott explains that such a study would likely find a very strong correlation between those who had recently visited a hospital and those who had died. Such an assessment could lead someone to think that visiting a hospital increases a person’s chance of untimely death. It is clear that this is not the case and such correlations should not be represented as causation. Likewise, the presence of a firearm in the house is unlikely to have caused the reported homicides. Without answering why these individuals owned firearms or whether or not these people were more likely to be attacked, it is impossible to fully determine the gun’s role in the crime.
Using the set of cases Kellermann looked at, it might be accurate to say that victims of homicides were more likely to have owned a firearm. Since homicide victims were Kellermann’s starting point, he could make such a claim. However, he cannot state that firearm owners are more likely to be victims of homicide. He did not obtain the random sample of gun owners that would have been necessary to test such a claim.
Lastly, Kellermann was extremely slow to release the data for his study. While we should not discount the study solely because of this, the fact that it took him 4 years to release a portion of the data is questionable. Such opacity substantially delayed effective analysis of his methodology and inhibited proper peer review at the time of its release.
While we should never be too quick to dismiss studies that produce undesirable results, we do need to critically review both pro- and anti-gun publications. In this case, the Kellermann study does not fare well against such examination.
Excerpt from More Guns, Less Crime by John Lott, Ph.D: http://crimepreventionresearchcenter.org/wp-content/uploads/2014/01/Screen-Shot-2014-01-15-at-Wednesday-January-15-11.17-AM.png
Gun Ownership as a Risk Factor for Homicide in the Home, Dr. Arthur Kellermann et al.: http://www.nejm.org/doi/full/10.1056/NEJM199310073291506