Kellermann's gun ownership studies after two decades

One method gun control advocates love to employ is tossing around a blizzard of studies in medical or other journals, claiming that owning a gun or carrying a gun is more of a danger to the owner than to some hypothetical bad guy.  The infamous example of this is epidemiologist Arthur Kellermann, beginning with his 1993 New England Journal of Medicine article, “Gun Ownership as a Risk Factor for Homicide in the Home.”  We’re informed that gun owners have a 2.7 times greater risk of homicide than those who don’t own guns.  I’ve known about this study for a while, but one of my readers, Esh325, reminded me that Kellermann’s claim needs to be addressed.

Others have criticized his specific techniques in analyzing the data, and I’ll leave those analyses to experts better able to do them.  In this article, my concern is about his initial assumptions, his conclusions from the data, and what factors he chooses to regard as significant.

Kellermann selected to study a case group of homicide victims, talking to a proxy—a relative of the victim—and a control group of people who lived in the same neighborhood but did not suffer the same crime.  These groups were from the Seattle, Memphis, and Cleveland metropolitan areas.  Some problems arise all along the way in this research.

At the time of the study, 1993, Seattle was the 52nd safest city out of 100, while Cleveland was 65th and Memphis was 74th.  Choosing cities in only the bottom half of safety rankings perhaps makes sense if you’re hunting for homicides, but as a study of the risks of gun ownership that we all face, this looks a lot more like cherry picking.  And while I’ll come back to the date of this study, let’s note that not much has changed, other than apparently getting worse.

The discussion at the end of Kellermann’s article acknowledges that homicide victims had a higher rate of domestic violence and substance abuse.  Many of us reading that are probably saying some variation of “No kidding!”  Should we be surprised that abusers and addicts have more bad outcomes in their lives?

This illustrates a flaw that shows up in a lot of these studies of the risks of gun ownership.  As I’ve noted before, factors such as drug or alcohol use, criminal behavior, and personal skill with firearms are often glossed over as though all gun owners are identical.  When a study like Kellermann’s starts off by comparing those who got murdered with those who didn’t, it introduces fatal flaws before things even get going.

Another flaw that people who quote (consciously or not) Kellermann suffer is the fact that his article was published in 1993.  America experienced a peak in violent crime in the 80s and early 90s, but those rates—including homicides—have been on the decline ever sinceGun homicides are at their lowest level in a long time.  More than 20 years on, enough has changed to call into question his assessment of danger incurred from owning firearms.

But what if we want to do valid research on the risks of owning guns?  The way to do it is with a longitudinal study.  That’s a long-term approach that takes a group of people—in this case, gun owners—and follows them over the course of decades.  This would allow us to see what percentage of the study group died due to guns—accidents, suicides, or homicides.  We could also learn how many of them use a firearm in self-defense and how many use their guns for other purposes like hunting or target shooting that benefit their lives.  Other data could be gathered—health, criminal behavior, education, and other factors that play into the stereotypes about gun owners.

Why isn’t that done?  As I said, such a study takes a long time.  It costs money.  But the data we could get would offer us valid science on the risks and benefits of gun ownership.

The views and opinions expressed in this post are those of the author’s and do not necessarily reflect the position of

Latest Reviews