Editorial: Deconstructing Kellermann

While a student many, many moons ago I worked as the secretary of a peer-reviewed scientific journal. Although my responsibilities were strictly clerical, I observed a thing or two about what happens behind the ivory walls of academia’s most sacred towers. While we prefer to think of all scientists as infallible servants of humanity, as indeed some are, many are corrupted by the same vices that plague, say, auto journalism: money, power, pride, and politics. Others are simply blinded by emotion. And yet many anti-gun advocates are quick to tout any study—no matter how flawed—as long as it agrees with their core beliefs. A recent post on Mikeb302000 featured this golden oldie: Gun Ownership As A Risk Factor For Homicide In The Home by Dr. Arthur L Kellermann, et al, circa 1993.  Let’s take a look…

The Kellermann study was held up by MikeB302000’s resident argumentum ad hominem attack dog and slayer of straw men, Jadegold. Seven times in her post and subsequent comments, she admonishes or criticizes gun rights proponents to read the oft-maligned study.

Kellermann, et al, conclude (twice in the abstract and three times in the discussion) that “keeping a gun in the home was strongly and independently associated with an increased risk of homicide (adjusted odds ratio, 2.7; 95 percent confidence interval, 1.6 to 4.4).” In other words, if you keep a gun in your home, you are 2.7 times more likely to have someone killed in your home—by that gun, by a gun introduced into your home by someone else, or other any other means.

Furthermore, they repeatedly assert that firearms are ineffectual in deterring homicide. “In light of these observations and our present findings, people should be strongly discouraged form keeping guns in their homes.”

These conclusions have been paraded by nearly every gun group in existence, from the Brady Bunch to Vegan Peace.  But they are wrong.

Flawed Methodology

For obvious reasons Kellermann, et al, could not interview homicide victims. So they had to rely on interviews of family members, roommates, neighbors or someone else deemed to be close to the victim. The researchers asked these proxies questions like “Did the victim keep a loaded gun in the house?” They also consulted police and autopsy reports.

For each homicide victim, Kellermann, et al, also randomly sought out someone else from the community who was demographically similar who had not died in a homicide to act as a control. In order to preserve consistency, the researchers did not interview the control subject directly; they selected a proxy for them as well. Again, in the interview, the control subject’s proxies were asked questions like, “Does the control subject keep a loaded gun in the house?”

The study showed that 45.4% of homicide victims kept a gun (any gun, loaded or unloaded) in the home. By comparison, only 35.8% of matching control subject proxies said that the control subjects kept guns in their homes. That’s far below more reliable national gun ownership statistics, including those touted by anti-gun groups.

The control percentage is underrepresented because proxies chosen for the control subjects aren’t necessarily privy to whether the control subject owns a gun and keeps it on the premises. For instance, none of my neighbors would possibly be able to testify as to whether I keep guns in my home. Nor do I know whether any of them keep guns. If you asked my father, mother, brother, or sisters, they wouldn’t be able to answer that question. The best they’d be able to offer is “maybe.”

As for the actual victims of homicide, there is little doubt as to whether they keep guns in the house because the police would have found them when they searched the premises. Or the coroner would have had to pry the gun from the victim’s cold dead fingers. [Sorry, I couldn’t resist.]

Kellermann, et al, concede that underreporting of gun ownership would skew up their numbers and blow up their conclusions. “We do not believe, however, that misreporting of gun ownership was a problem,” they write, because a separate pilot study confirmed their results against handgun registrations and other surveys.

But what about all of the guns not registered? Did they not think there might be more than a few Saturday Night Specials tucked in the waistbands of residents in the high-crime neighborhoods they selected for their study?

Kellermann’s methodology for counting firearm ownership among their control subjects was flawed. The number is low, which means that the math that computes an adjusted odds ratio of 2.7 is way off. Indeed, if you substitute the national average of 48% for the control group, it would show that having a gun in the home was a negative risk factor for homicide (i.e. lessened the chances of homicide in your home).

Guns Not Effective to Deter Homicide?

The certain conclusion Kellermann, et al, make that firearms in the home “confer protection” is wholly unsupported because defensive uses of firearms were only counted in this study if they resulted in a homicide. Of the 420 homicides included in the study, 21 homicide victims [unsuccessfully] attempted to use a gun for self-defense. In only 15 instances were bad guys killed under “legally excusable circumstances,” meaning they were shot by the police (4) or by someone acting in self-defense (11).

According to Kellermann, et al, you haven’t successfully defended yourself from homicide unless you killed the bad guy dead. If the residents canvassed in this study fired a warning shot that chases off a bad guy, it didn’t count. If they wounded a perp, it didn’t count. And if they brandished their weapon and scared a threat off, Kellermann, et al, didn’t count it.

By their measure, the only good bad guy is a dead bad guy. As such, their definition of a successful use of a firearm in self-defense is wholly insufficient to conclude that firearms are ineffectual for self-protection.

Common Sense: More than a Pamphlet by Paine

Despite the flawed adjusted odds ratio, the Kellermann study actually places guns kept in the home fifth (2.7 ratio) among six risk variables for homicide in the home. The greatest risk factor they identified was actually whether anyone in the household used illicit drugs (5.7 ratio), followed by whether the home was rented and if there was a history of domestic violence (both 4.4 ratios). Even living alone (3.7 ratio) was a greater risk factor than the presence of a gun.  If the percentage of gun ownership among the control group was corrected, gun ownership would fall even further down the list.

Nonetheless, Dr. Arthur L. Kellermann and the anti-gun Nazis continue to dredge up this study as evidence that “guns kill.” Yes they do, but not without a hand to guide them. Don’t believe everything you read about who, when, where and how that happens.