How to Count the Defensive Use of Guns

Neither survey calls nor media and police reports capture the importance of private gun ownership.


Stringent gun control advocates are fond of underestimating the possible importance of owning a gun. For example, a pair of anti-gun activists took to Politico in January to claim that the gun rights community is deluded about the likely number of defensive uses of guns by American citizens. Such defensive uses are known as DGUs ("defensive gun uses") in the lingo.

Many in the gun rights community believe that a privately owned gun is used in legitimate self-defense over 2 million times a year in America. This figure arose initially from the survey work done in 1993 by Florida State University criminologists Gary Kleck and Marc Gertz.

That study, known as the National Self-Defense Survey (NSDS), involved conducting nearly 5,000 interviews by phone in 48 states, anonymously, with phone numbers randomly generated. The respondents were asked if over the past five years they had used a gun—even if not fired—for self-protection or protection of property at work, home, or elsewhere. They were asked to exclude any use in miitary, police, or security work, if applicable, and asked whether the gun use was against an animal or a person.

If their response indicated a DGU, they were asked if the gun use had occurred in the past year, and if it was the speaker or another member of the household who had the experience. Those who claimed DGUs were then asked what Kleck described in his 2001 book Armed as a "detailed series of questions establishing exactly what happened" in the incidents. 

Kleck says they found 222 bonafide DGUs directly via survey. The defender had to "state a specific crime they thought was being committed" and to have actually used the gun, even if just threateningly or by "verbally referring to the gun." Kleck insists the surveyors were scrupulous about eliminating any sketchy or questionable-seeming responses.

The study concluded, based only on stories said to have occurred to the speaker during the past year, and extrapolating from their results, that 2.2 to 2.5 million DGUs happened in the U.S. a year.

Kleck points out this only means that about 1 percent of guns in the U.S. are thus used annually. In Armed, Kleck discusses a number of later surveys on DGUs, including one from the federal Centers for Disease Control and Prevention in 1994, that at least roughly back up his original estimates. He sums up that "there are now at least nineteen professional surveys, seventeen of them national in scope, that indicate huge numbers of defensive gun uses in the U.S."

The one huge outlier in finding so many DGUs is the National Crime Victimization Survey (NCVS), also done 1993-94 by the Census Bureau on behalf of the Department of Justice, with a follow-up in 1998. That one found just 116,000—a near 95 percent drop from the NSDS survey result. The NCVS was designed to gather information, as the title indicates, on crime victimization in general with the DGU response a byproduct.

Kleck has suggested some reasons for the low NCVS number that don't rely on its DGU estimate actually being correct. Among them are that NCVS never explicitly asks about DGUs, merely asking those who say yes to having been a crime victim whether they "did or tried to do" something about it while it was happening; that people might be reluctant to admit to possibly criminal action on their own part (especially since the vast majority of crime victimizations in the survey occurred outside the home, where the possession's legality might be questionable) to a government surveyor after they've given their name and address (which they were asked to do); that they might consider a crime prevented by a DGU as not a crime they were victims of at all since it was uncompleted; and that there are some independent reasons to believe the very crime victimization numbers, which the DGUs are a subset of, are undercounted by the survey. 

The article in Politico by Evan Defilippis and Devin Hughes directly took on Kleck's numbers and those who repeat them. The authors, both with the anti-gun group Armed With Reason, consider the numbers myths and those who believe and repeat them mythmongers.

The issues they raise with Kleck's work have floated around the literature ever since Kleck first published them. In rough summation, Defilippis and Hughes insist that lots of people undoubtedly lied to Kleck and his researchers, and similarly lied to any other survey researchers who reached similar conclusions. Why would they lie in such vast numbers? Because they were gun owners who wanted to justify their ownership of guns, justify the social value of guns for ideological reasons, wanted to appear heroic to the anonymous surveyors, or were mistakenly "telescoping" events that happened longer than a year ago into the one-year estimate.

It is not enough for them to declare that Kleck's survey's don't prove what he and his fans think. Defilippis and Hughes simultaneously promote a couple of much lower estimations of DGUs as the real truth, citing two data sources that, rather than surveying, simply count verified reports of DGUs in the media and in police reports.

One study which they highlight was from Arizona in 2004, published in the journal Injury Prevention. (They don't highlight that the study is over a decade old.) The study was done by Arizona State University psychology professor W.V. Fabricius and an Arizona high school teacher, J.F. Denton. Their method? "Reported uses of firearms in a newspaper covering roughly the Phoenix metropolitan area over almost a 3.5 month period were examined, supplemented where necessary by police and court records." They found three cases where Kleck/Gertz's conclusions should have found over 300. 

Only cases that had been reported to the police or to a newspaper would end up in their report. Self-reflection on the part of any armed citizen who can imagine themselves having had to brandish or use their weapon would show it isn't necessarily in your best interest to call police attention to the matter.

After all, your possession or use of the weapon might be a matter of greater concern to the cops than whatever the intruder or criminal you were repelling was up to. They'll doubtless never lay hands on him; you are right there, for any investigation and harassment the cops might want to call forth. Many gun owners or gun users might see little good and much possible bad arising from calling the cops after a DGU incident, and thus many or even most would never make a police blotter, never make a newspaper.

Kleck is confident even his initial estimates almost certainly understate DGUs. He believes even his anonymous pollsters might not have won the trust of everyone they talked to, that people may have discounted some incidents as unimportant, that some may have involved people under 18 (who were not polled), and that phoneless (usually poorer) households may be overrepresented among DGUs while obviously represented not at all in this phone poll.

Fabricius and Denton admit, at the end of their three-page Injury Prevention paper, that if the gun was in fact not fired in the course of the DGU, then indeed neither police nor media would likely report it. They don't note that Kleck/Gertz's study found that 76 percent of the DGUs did not involve firing the weapon.

They don't try to account for this or assume that perhaps these cases explain part or all of the huge disconnect between what their counting method and Kleck's counting method found. They do, just in the name of objective social science, of course, make the sideways comment that if that's true, then no one needs a loaded gun anyway if merely waving one is good enough.

Politico's other source to cast shade on Kleck is at least more current: the site of the Gun Violence Archive. They get their counts from "automated queries, manual research through over 1,200 media sources, aggregates, police blotters, police media outlets and other sources daily."  Their 2014 figure for what they consider verified DGUs is 1,581, with 204 so far in 2015. (Recall we have very little reason to believe that many DGUs would ever end up in any objectively verifiable outside source or that the newspaper/police report method would come anywhere near to an accurate count of DGUs.)

A paper by Clayton Cramer and David Burnett called Tough Targets, issued by the Cato Institute, does a good layman's survey of all the extant survey evidence regarding DGUs and concludes there's at least fair reason to believe both high and low estimates aren't precise. Like the Gun Violence Archive, they came up with their own set of numbers and case studies about DGUs by studying newspaper reports—in their case eight years worth nationally via the Armed Citizen reports from the NRA amounting to nearly 5,000 incidents.

They sum up some of the trends and tell some of the specific stories to provide a more wide-ranging and nuanced set of some of the circumstances and reasons people use guns defensively in a manner that makes the news. Studying them should remind people that raw numbers of people who have to use guns defensively aren't particularly relevant to any citizen who has to use a gun defensively.

Some interesting details include 25 rapists challenged with weapons, 154 women defending themselves, only nine involving drug dealers, more criminals having their guns taken by victims than the oft-feared other way around, and 34 pizza deliverers having to defend themselves against crime with a gun.

To the people involved in these incidents, no number of DGUs is more important than the cardinal number representing them.

Kleck rose to his own defense last month in Politico, arguing that the earlier piece repeated critiques originally and more famously brought up all the way back in 1997 by David Hemenway. Kleck has already addressed them, in particular detail in his book Armed

Very briefly, Kleck questions the method of those not versed in survey research, like the Politico authors and Hemenway, of just making wild guesses about all the places the surveyors must have gone wrong, if the surveyors' results don't match the critics mere intuitions. (Certainly, nothing is inherently implausible about the idea that 1 percent of guns in the U.S. might be used a year defensively.)

The nub of Kleck's argument against his critics is that, sure, one can make guesses about reasons why surveys might overcount, but experienced surveyors know that research "consistently indicates that survey respondents underreport (1) crime victimization experiences, (2) gun ownership and (3) their own illegal behavior."


He also notes that while false positives for DGUs from surveys are certainly possible, "false negatives…could be (and, according to extensive research, are) even more common. In that case, survey estimates of DGU frequency would be too low, not the enormous overestimate that DeFillipis and Hughes believe in." Critics of his DGU research, Kleck writes, never have "made the slightest effort to estimate the number of false negatives, they cannot possibly know whether false positives outnumber false negatives and therefore have no logical foundation whatsoever for their claims that erroneous responses to DGU questions result in an overestimate of DGU frequency."

So, how many DGUs are there, really? My conclusion is the same is it was when I wrote my 2008 book Gun Control on Trial: There isn't any real way to know for sure. Some of the anomalies in the basic Kleck/Gertz numbers make one wonder if they are reliably accurate, including the small number of gun-wounded folk showing up in hospitals compared to how many might be expected from the over-2-million-DGU figure, or how many burglaries there are compared to the number that people claim to have used guns to defend themselves from them, and a difficult-to-believe large number of women who seem to have DGUs from the Kleck/Gertz numbers when compared to the much smaller percentages of women who own guns or are involved in recorded justifiable homicides.

In a 1997 paper for the Journal of Criminal Law and Criminology, Tom Smith makes some reasonable though unverified guesses about ways both the NSDS and NCVS survey might be expected to over- and underestimate DGUs and is satisfied he is able to adjust an over 30:1 difference to a mere 3-5:1 one. Certainly, any data that ultimately depends on people accurately remembering and relating stories about such an emotionally and legally charged issue with time-accuracy are bound to have some mistakes baked in, minus some absolutely objective other measure that shows the same thing. (And no, newspaper and police reports come nowhere close to being such an objective measure, even though they don't show the same thing.) Those mistakes may balance out, or they may not.

Curiosity about any fact of reality is a good trait. It helps, though, to be empirical and scientific enough to know that certain numbers are beyond our precise knowledge. As hard as it is, it's also intellectually honest to be as skeptical about the educated guesses that you think support your political conclusion as you are about ones that oppose it.

Knowing for sure how many DGUs there are in America in any given year is irrelevant, however, to any important political, ethical, or certainly constitutional, question about gun ownership in America.

Guns are tools that exist in the world. They cannot be fantasized or legislated away, neither constitutionally nor actually. They can be used to harm the innocent or guilty, and they can be used to defend the innocent or guilty. And no number of other people misusing them, or even better never using, their weapons has any bearing on whether you should be able to have one if you think you need it.

Those people who lived out the stories in any case study collection of newspaper or police reports of DGUs would doubtless find it curious to hear they shouldn't have had the right to defend themselves, because an insufficiently impressive number of other citizens had done the same. But underestimating the significance of what's at stake in Second Amendment rights—even though it can clearly be life itself, not to mention dignity—is a favorite pastime of gun controllers and their ideological soldiers.

The right to self-defense is a core right of human nature. It is so core that those who argue for the necessity of government generally claim that its supplying of defense is one of the reasons we can't live without it. But every reasonable person knows that in no case of an intruder in one's home or outside of it intending to harm you is the state's police going to do a thing to help you.

As I concluded in my 2008 book Gun Control on Trial:

The opposing armies in the DGU war are roughly staked out with these dueling positions: 1) "There are a really large number of defensive gun uses, so many that any reasonable person would have to admit that private gun ownership is some kind of social good" and 2) "While there may be a fair number of DGUs, the number is dwarfed by the number of violent crimes committed with guns, so never mind the people who save themselves with guns, we should let politicians concentrate not on speculative and uncertain defensive uses, but on the crimes and loss of life and limb that we can see and count which accompanies gun possession and use."

Left out of any policy decision based on these sorts of macrostatistics, as always, is how much having a gun mattered to the specific individual person able to defend himself.

However large the number of DGUs, or how small; and however large the number of accidents or tragedies caused by guns, or how small, the right and ability to choose for yourself how to defend yourself and your family—at home or away from it—remains, and that numerical debate should have no particular bearing on it.