Some papers in the medical literature have written a homeowner's gun is more likely to kill its owner or family member than kill a criminal, and therefore "the advisability of keeping firearms in the home for protection must be questioned." The most notable (or notorious), and quoted in the previous sentence, is written by doctors Arthur Kellermann and Don Reay, and is titled, "Protection or peril? An analysis of firearms related deaths in the home." (New Engl J Med 1986. 314: 1557-60.)
The oft cited Kellermann paper found a homeowner's gun was 43 times more likely to kill a family member, friend, or acquaintence, than it was used to kill someone in self-defense. Kellermann stated, "for every case of self-protection homicide involving a firearm kept in the home, there were 1.3 accidental deaths, 4.6 criminal homicides, and 37 suicides involving firearms." Florida State University professor Gary Kleck appropriately terms these ratios "nonsensical." (Targeting Guns: Firearms and Their Control, pp. 177-179, 1997)
Although this study was published in 1986 its findings continue to be uncritically cited in medical journals, government publications, and non-technical periodicals such as health newsletters, general interest magazines, op-ed pieces, letters-to-the editor, etc.
Not only is Kellermann's methodology flawed, but using the same approach for violent deaths in the home not involving a firearm, the risk factor more than doubles from 43 to 1, to 99 to 1. Let's see why this 43 to 1 ratio is a meaningless indicator of gun ownership risk.
First we need to understand how the ratio was derived.
Kellermann tabulated gunshot deaths occurring in King County, Washington, from 1978 to 1983. Table 1 below is taken from Kellermann's paper (Table 3 on p. 1559).
Table 1. Classification of 398 Gunshot Deaths involving a Firearm Kept in the Home
Type of Death No. Unintentional deaths 12 Criminal homicide 41 Suicide 333 Unknown 3 Total 389 Self-protection homicide 9
As we see from Table 1, a ratio of 389 violent deaths to 9 justifiable homicides gives us the famous 43 to 1 ratio.
Let's apply the same methodology to non-gun deaths and non-gun self-protection homicides in the home, for King County, Washington.
Table 2. Estimation of Violent Deaths in the Home Not Involving a Firearm
Type of Death No. Unintentional deaths 0 Criminal homicide1 50 Suicide2 347 Unknown 0 Total 397 Self-protection homicide3 4
This ratio of 397 non-gun violent deaths to 4 justifiable homicides reduces to 99 to 1.
So having applied Kellermann's methodology to non-firearm violent death, the risk factor more than doubles from 43 to 1, to 99 to 1.
Please note, the purpose of this exercise is not to show that using a gun in the home is better than not using one. This exercise does no such thing. It is merely to show how deeply flawed Kellermann's study really is. Further, a number of tremendously important factors are left unaccounted.
For example, another way of looking at it is, more martial artists are probably murdered by non-gun methods than they kill in self-defense. Would we conclude that it is best to avoid learning a martial art for self-defense based on such a "nonsense ratio?" Regardless of how the number crunching had turned-out between gun and non-gun violent deaths in the home, we should be able to see that Kellermann's approach contributes nothing towards establishing a general or personal risk factor for a gun in the home.
What is truly sad about the nonsense-ratio is how often it is cited and uncritically accepted.
To decide whether or not to own a gun for self-defense based solely on a "kill" ratio is folly. To estimate the risks and benefits of gun ownership many more factors need to be considered. An example is defensive gun use, which outnumbers homicides, suicides, and accidents, and is ignored in most of the medical research. (See How often are guns used in self-defense?)
For a different approach in critique of Kellermann's study see The 43: 1 Fallacy by Dave Kopel.
For Further Reading
GunCite's critique of Kellermann's "3:1" study. More generally, see GunCite's Gun Control Research.
A criticism of Kellermann's subsequent research, and the bias of the Center for Disease Control's firearm related research: Kates, Schaffer, and Waters, Public Health Pot Shots: How the CDC Succumbed to the Gun "Epidemic", Reason Magazine, April 1997.
Scroll down to part part XV:"Gun Ownership as a Risk Factor for Homicide in the Home": Kates, Schafer, et. al, Guns and Public Health: Epidemic of Violence or Pandemic of Propaganda?. Originally published as 61 Tenn. L. Rev. 513-596 (1994).
Letters to the New England Journal of Medicine regarding Kellerman's paper titled: "Guns and Homicide in the Home".
Kleck, Gary, What Are the Risks and Benefits of Keeping a Gun in the Home?, JAMA, August 5, 1998.
A differing view from Kleck's: Peter Cummings; Thomas D. Koepsell, Does Owning a Firearm Increase or Decrease the Risk of Death?, JAMA, August 5, 1998.
Letter to the editor and a response from Kleck, JAMA, July 14, 1999.
1. Non-gun criminal homicide calculation:According to Kellermann, firearms were involved in 45 percent of all homicides in King County.2. Non-gun suicide calculation:
41 firearm criminal homicides / .45 = 91 total criminal homicides.
Non-gun criminal homicides = 91 / (1 - .45) = 50 non-gun criminal homicides.According to Kellermann, firearms were involved in 49 percent of all suicides in King County.3. Self-protection calculation:
333 firearm suicides / .49 = 680 total suicides.
Non-gun suicides = 680 / (1 - .49) = 347 non-gun suicides.According to the 1997 FBI Uniform Crime Report (p. 24), from 1993 to 1997, non-gun justifiable homicides were 13% of all justifiable homicides. 30% was used instead of 13%.
9 firearm justifiable homicides / (1 - .3) = 13 total justifiable homicides.
13 total justifiable homicides - 9 firearm justifiable homicides = 4 non-gun justifiable homicides.