Guns and Self-Defense

by Gary Kleck, Ph.D.

This afterword [to J. Neil Schulman's book, Stopping Power] is abridged from chapter five of Targeting Guns: Firearms and Their Control Copyright © 1997 by Walter de Gruyter, Inc., New York. Used by permission of the publisher, Aldine de Gruyter, and the author.

Citations have been deleted and the original chapter should be consulted for all references.

-------------

Gun ownership for self-protection, and defensive gun use, must be distinguished from other forms of forceful activity directed at criminals, such as private vigilantism, or the activities of the criminal justice system, such as police making arrests. All of these can be coercive and all may be done by armed persons. However, vigilantism and criminal-justice activity share a purpose that self-defensive actions do not—retribution. Whereas the criminal-justice system and the vigilante both seek to punish wrong-doers, the defensive gun user seeks to protect the bodily safety and property of himself and others. Retribution is neither an essential nor even necessarily common part of self-defense actions. Further, the vigilante proactively seeks out contact with criminals, while the defender typically reacts to actions initiated by criminals. The true vigilante acts collectively, in concert with like-minded individuals, whereas the defender ordinarily acts alone. It therefore is an oxymoron to refer to a defensive gun user as a “lone vigilante.” Further, gun ownership is largely passive self-protection—once a gun is acquired, the owner only rarely does anything defensive with it. Only a minority of defensive owners actually use their guns for self-protection; most of the rest just keep the gun in a bureau drawer or similar location, where it is available for use should the need arise. This contrasts sharply with neighborhood crime control strategies, that may require considerable investment of time and effort from each participant.

Although gun ownership costs more money than simple measures such as locking doors, having neighbors watch one’s house, or avoidance behaviors such as not going out at night, it costs less than buying and maintaining a dog, paying a security guard, buying a burglar alarm system, or relocating one’s residence to an area with less crime. Consequently, it is a self-protection measure available to many low-income people who cannot afford more expensive alternatives. Gun ownership is not a replacement or substitute for these other self-help measures, nor for criminal-justice-system activities, but rather is more accurately thought of as a complement to them—an additional measure that might prove useful, for at least some crime victims, some of the time.

At least 12 national and 3 state-wide surveys have asked probability samples of the general adult population about defensive gun use. The surveys differ in many important respects. The two most sophisticated national surveys are the National Self-Defense Survey done by Marc Gertz and myself in 1995 and a smaller scale survey done by the Police Foundation in 1996.

The National Self-Defense Survey was the first survey specifically designed to estimate the frequency of defensive gun uses. It asked all respondents about both their own uses and those of other household members, inquired about all gun types, excluded uses against animals or connected with occupational duties, and limited recall periods to one and five years. Equally importantly, it established, with detailed questioning, whether persons claiming a defensive gun use had actually confronted an adversary (as distinct from, say, merely investigating a suspicious noise in the backyard), actually used their guns in some way, such as, at minimum, threatening their adversaries (as distinct from merely owning or carrying a gun for defensive reasons), and had done so in connection with what they regarded as a specific crime being committed against them.

The National Self-Defense Survey indicated that there were 2.5 million incidents of defensive gun use per year in the U.S. during the 1988-1993 period. This is probably a conservative estimate, for two reasons. First, cases of respondents intentionally withholding reports of genuine defensive-gun uses were probably more common than cases of respondents reporting incidents that did not occur or that were not genuinely defensive. Second, the survey covered only adults age 18 and older, thereby excluding all defensive gun uses involving adolescents, the age group most likely to suffer a violent victimization.

The authors concluded that defensive uses of guns are about three to four times as common as criminal uses of guns.

The National Self-Defense Survey confirmed the picture of frequent defensive gun use implied by the results of earlier, less sophisticated surveys.

A national survey conducted in 1994 by the Police Foundation and sponsored by the National Institute of Justice almost exactly confirmed the estimates from the National Self-Defense Survey. This survey’s person-based estimate was that 1.44% of the adult population had used a gun for protection against a person in the previous year, implying 2.73 million defensive gun users. These results were well within sampling error of the corresponding 1.33% and 2.55 million estimates produced by the National Self-Defense Survey.

The one survey that is clearly not suitable for estimating the total number of defensive gun uses is the National Crime Victimization Survey. This is the only survey that has ever generated results implying an annual defensive-gun-use estimate under 700,000. Not surprisingly, it is a favorite of academic gun-control supporters. If one is to make even a pretense of empirically supporting the claim that defensive gun use is rare in America, one must rely on the National Crime Victimization Survey, warts and all.

That the National Crime Victimization Survey estimate is radically wrong is now beyond serious dispute. Ultimately, the only foundation one ever has for knowing that a measurement is wrong is that it is inconsistent with other measurements of the same phenomenon. There are now at least 15 other independent estimates of the frequency of defensive gun uses and every one of them is enormously larger than the National-Crime-Victimization-Survey estimate. Unanimity is rare in studies of crime, but this is one of those rare cases. Apparently, however, even unanimous and overwhelming evidence is not sufficient to dissuade the gun control advocacy organizations, such as Handgun Control, Inc., and the Coalition to Stop Gun Violence, that the National Crime Victimization Survey estimate is at least approximately valid and that defensive gun use is rare.

The numerous surveys yielding contrary estimates strongly support the view that the National-Crime-Victimization-Survey estimate is grossly erroneous.

There has probably been more outright dishonesty in addressing the issue of the frequency of defensive gun use than any other issue in the gun control debate. Faced with a huge body of evidence contradicting their low defensive-gun-use position, hard-core gun-control supporters have had little choice but to simply promote the unsuitable National-Crime-Victimization-Survey estimate and ignore or discount everything else. Authors writing in medical and public health journals are typically the most crudely dishonest—-they simply withhold from their readers the very existence of a mountain of contradictory evidence.

Adherents of the rare-defensive-gun-use thesis also use another tactic, in addition to simply pretending that the contrary evidence does not exist. On those rare occasions when they briefly and very partially address some of the contrary evidence, they counter evidence with one-sided speculation rather than better empirical information.

Even if some of these speculations had been correct and consequential, it is not productive or legitimate to speculate only in one direction, in this case speculating only about flaws that supposedly pushed defensive-gun-use estimates up. If one is not willing to seriously consider errors in both directions, one is simply engaging in “adversary scholarship” or “sagecraft,” an enterprise aimed not at discovering the truth, but rather at buttressing predetermined positions.

Speculation about the flaws in surveys estimating large numbers of defensive gun uses resemble UFO buffs’ beliefs that the federal government captured aliens from other worlds at Roswell, New Mexico in 1947. The reason most people do not share these beliefs is not that they can be proven false; they cannot, since it is impossible to prove a negative. Rather, most people reject them because there is no credible evidence that they are true. It is the same with speculations about the gun surveys’ supposed flaws. Since it is impossible to prove a negative, one cannot prove that massive misreporting of nonexistent defensive gun use incidents did not occur in the gun surveys. There is, however, no evidence whatsoever that such massive misreporting did occur.

Faced with such overwhelming survey support for the idea that defensive gun uses are common, some pro-control scholars belatedly adopted the view that surveys simply cannot yield useful information about how often they occur. Faced with defeat on the field of empirical evidence, they suddenly developed a radical skepticism toward all survey estimates.

Most uses of guns for either criminal or defensive purposes are less dramatic or consequential than one might think. Only 3% of criminal gun assaults involves anyone actually being wounded, even nonfatally, and the same is true of defensive gun uses. More commonly, guns are merely pointed at another person, or perhaps only referred to (“I’ve got a gun”) or displayed, and this is sufficient to accomplish the ends of the user, whether criminal or noncriminal. Nevertheless, most gun owners questioned in surveys assert that they would be willing to shoot criminals under the right circumstances. A 1989 Time/CNN survey found that 78% of gun owners said they would shoot a burglar if they felt threatened by that person.

Despite this stated willingness of gun owners to shoot under certain circumstances, most defensive uses of guns do not in fact involve shooting anyone. Data from the National Self-Defense Survey indicate that no more than 8% of the 2.5 million annual defensive gun uses involved a defender who claimed to have shot their adversaries, or about 200,000 total. The 8% figure, however, should be taken with a grain of salt because it is based on a sample of only 213 cases, 17 of whom reported a wounding, and because the respondants were not asked how they knew they had wounded the criminals. In cases where the criminal escaped, these reports may often have been based on favorable guesses about the shooter’s marksmanship skills. As Marc Gertz and I noted, the claimed “hit rate” of shooters in the National-Self-Defense-Survey-reported incidents was higher than that of police officers, an unlikely level of shooting skill under stress.

Regardless, there is nothing even mildly implausible about 200,000 annual nonfatal woundings linked with defensive gun uses, in light of the estimated 150,000 annual medically treated gunshot woundings. There could also easily be just as many more that went untreated because they involved criminals wounded by their victims during a criminal attempt. Criminals wounded in the course of attempting a crime against an armed victim are almost certainly a set of persons lying largely outside the set of persons receiving medical treatment for gun-shot wounds, and thus the size of the latter set can tell us nothing about how large the former should be.

The rarest, but most serious form of self-defense with a gun is a defensive killing. The FBI does not publish statistics on self-defense killings per se, but it did start publishing counts of civilian justifiable homicides gathered through their Supplementary Homicides Reports program in their 1991 issue. For a variety of reasons, the FBI counts of civilian justifiable homicides represent only a minority of all civilian legal defensive homicides. FBI-counted civilian justifiable homicides were used to estimate total civilian legal defensive homicides. FBI counts of police justifiable homicides are also reported here. Regardless of which counts of homicides by police are used, the results indicate that civilians legally kill far more felons than police officers do. The figures imply that, of 24,614 civilian (not by police) homicide deaths in the United States in 1990, about 1400 to 3200, or 5.6% to 13.0% were legal civilian defensive homicides.

This estimate was independently confirmed by the only national study of homicide dispositions done to date. Analysis of 231 homicides occurring in the U.S. in the first week of May, 1989 indicated that between 15 (6.5%) and 28 (12.1%) were ruled justifiable.

Nonfatal gun woundings are far more frequent than fatal shootings. In 1985 Cook reviewed data that indicate that about 15% of assault-linked gunshot wounds known to the police are fatal, implying a ratio of about 5.67 (85/15) reported nonfatal assaultive gun woundings to each fatal one. Assuming the same applies to legal civilian defensive shootings, there were between 6,300 and 15,300 reported nonfatal, legally permissible woundings of criminals by gun-armed civilians in 1990. Combining the defensive killings and nonfatal woundings, there are about 7,700 to 18,500 reported legal shootings of criminals a year, which would be less than 1% of all defensive gun uses. The rest of defensive gun uses, then, involve neither killings nor woundings but rather misses, warning shots fired, or guns used to threaten, by pointing them or verbally referring to them.

That defensive gun uses, with or without a wounding, are so common is not surprising in view of how many Americans own guns for defensive reasons and keep them ready for defensive use. A 1989 national survey found that 27% of gun owners have a gun mainly for protection, and 62% said that protection from crime was at least one of the reasons they owned guns. This translates into about 16 million people in 1993 who had guns mainly for protection, and about 36 million in 1993 who had them at least partly for protection.

Further, many gun owners, and almost certainly a majority of those who own guns primarily for protection, keep a household gun loaded. The 1989 survey found that 24% of gun owners always keep a gun loaded, and another 7% had a gun loaded at the time of the interview although they did not do so all the time, for a total of 31%. Guns were most commonly kept in the bedroom, where they would be ready for nighttime use.

Are gun defenders really vengeful vigilantes, seeking out criminals to inflict punishment on them? If this were true, we would expect defenders to express more punitive views towards criminals. Results from the National Self-Defense Survey did not support this imagery. Both with regard to support for the death penalty and opinions concerning whether the courts are sufficiently harsh towards criminals, gun defenders were essentially identical to nondefenders. Instead, gun defenders were drawn from the ranks of groups with higher-than-average risks of criminal victimization: males, young adults, minorities, and big-city dwellers. And, of course, gun defenders were more likely to own guns and to carry them for protection.

Before addressing the objective effects of actual defensive uses of guns, a more subjective issue should be addressed. If some people get guns in response to crime or the prospect of being victimized in the future, does a gun make its owner feel safer?

A December 1989 CNN/Time national survey of 605 U.S. gun owners asked the following question: “Does having a gun in your house make you feel more safe from crime, less safe, or doesn’t it make any difference?” Of the gun owners, 42% felt more safe, 2% felt less safe, and the rest said it made no difference. Results were virtually identical in a May 1994 survey for U.S. News and World Report. When asked “Overall, do you feel comfortable with a gun in your house or are you sometimes afraid of it?,” 92% of gun owners said they were comfortable, 6% were sometimes afraid, and 2% were not sure.

In sum, most gun owners, including many who do not even have a gun for defensive reasons, feel comfortable with guns, feel safer from crime because of them, and believe their guns actually do make them safer.

Another way of approaching this issue is to ask people how they would feel if guns were eliminated. If widespread gun ownership was currently making people feel less safe, then eliminating guns should make them feel more safe. An August, 1994 Gallup poll asked: “Suppose a law were passed which you were certain would remove all handguns from the possession of all citizens other than the police. Would you feel more safe, less safe, or wouldn’t it make any difference?” While 32% said they would feel more safe, 41% said they would feel less safe, and the remainder felt it would make no difference. Since there are more who would feel less safe than who would feel more safe, the net effect on the population as a whole of eliminating guns would be to make the population feel less safe.

Do gun owners’ feelings of greater security, however real in emotional terms, have any factual foundation? Even proponents of stringent gun control who assert that guns are not effective defensive devices for civilians nearly always make exceptions for police officers and the like. The rationale for police having guns is based at least partly on the idea that police need and can effectively use guns for defending themselves and others. Doubts about the defensive utility of guns, then, appear to rest on any of three beliefs: (1) civilians do not need any self-protective devices, because they will almost never confront criminals, or at least will almost never do so while they have access to a gun, or (2) they do not need guns because they can rely on the police for protection, or (3) civilians, unlike police, are not able to use guns effectively, regardless of need.

Regarding the first belief, National Crime Victimization Survey estimates indicate that 83% of Americans will, sometime over the span of their lives, be a victim of a violent crime, all of which by definition involve direct confrontation with a criminal. Although it cannot be stated what share of these incidents will transpire in a way that would allow the victim to actually use a gun, it is clear that a large share of the population will experience a violent victimization.

The second idea, that citizens can depend on police for effective protection, is clearly erroneous. It implies that police can serve the same function as a gun in disrupting a crime in progress, before the victim is hurt or loses property. Police cannot do this, and indeed do not themselves even claim to be able to do so. Instead, police primarily respond to crimes after they have occurred, questioning the victim and other witnesses in the hope that they can apprehend the criminals, make them available for prosecution and punishment, and thereby deter other criminals from attempting crimes. Police officers rarely disrupt violent crimes or burglaries in progress; even the most professional and efficient urban police forces rarely can reach the scene of a crime soon enough to catch the criminal “in the act.”

The third idea, that civilians are not generally able to use guns effectively, requires more extended consideration. Gun-control proponents sometimes argue that wielding guns effectively in self-defense requires special training, skills, and emotional control that only police have. They hint that would-be gun users are ineffectual, panic-prone hysterics, as likely to accidentally shoot a family member as a burglar, occasionally citing a few illustrative anecdotes for support.

Incidents in which householders shoot family members mistaken for burglars and other criminals have occurred, but they are extremely rare. Studies indicate that fewer than 2% of fatal gun accidents involve a person accidentally shooting someone mistaken for an intruder. With 1409 fatal gun accidents in 1992, this implies that there are fewer than 28 incidents of this sort annually. Compared with about 2.5 million annual defensive uses of guns, this translates into about a less than 1-in-90,000 chance of a defensive gun use resulting in this kind of accident.

It has also been claimed that many people who attempt to use guns for self-protection have the gun taken from them by the criminal and used against them. Although this type of incident is not totally unknown, it too is extremely rare. In the 1979-1985 National-Crime-Victimization-Survey sample, it was possible to identify crime incidents in which the victim used a gun for protection and lost a gun to the offender(s). At most, 1% of defensive gun uses resulted in the offender taking a gun away from the victim. Even these few cases did not necessarily involve the offender snatching a gun out of the victim’s hands. Instead a burglar might, for example, have been leaving a home with one of the household’s guns when a resident attempted to stop him using another household gun. Thus, the 1% figure probably represents an upper limit.

Some people have misinterpreted data on police killings as indicating that even trained police officers can have their guns taken away while attempting to use them for self-defense. With about 600,000 sworn officers carrying guns on duty, and an average of about 68 killed per year, about eight of which per year were killed with their own weapons, this would imply an annual risk of 1.4 police officers killed with their own guns per 100,000 officers.

More misleading still was the suggestion that these rare killings bore on the issue of the risks of defensive gun use. In a detailed study of killings of officers, the FBI found that, among 11 officers who were killed with their own guns, only one involved a gun taken from the victim officer’s hand. Since actually using a gun for self-defense would have to involve an officer holding the gun, this implies that cases of officers killed with their own guns almost never involve an attempt by the victim officer to use the gun for self-defense. Instead guns were typically taken from the officer’s holster or vehicle. Police officers are almost never killed by offenders who took their guns away while the officers were trying to use the guns defensively. Police killings therefore offer no support for the notion that using guns for self-defense is risky.

Based on nationally representative samples of crime incidents reported in the National Crime Victimization Surveys, victims who use guns for self-protection were less likely to be injured or to lose property than otherwise similar victims who used other forms of self-protection or who did not resist at all. For example, among robbery victims who used guns, only 17% were injured and only 31% lost property, compared to 25% injury rates and 88% property loss rates among victims who did not resist at all, and 33% injury rates and 65% property loss rates among all robbery victims.

Some police officers advise people to refrain from armed resistance should they be confronted by a criminal. For example, Joseph McNamara, then the Chief of the San Jose Police Department, testified before a Congressional committee: “We urge citizens not to resist armed robbery, but in these sad cases I described, the victims ended up dead because they produced their own handguns and escalated the violence. Very rarely have I seen cases in which the handgun was used to ward off a criminal.”

Given the foregoing evidence, McNamara’s factual premises are clearly wrong.

Why, then, do some police give such advice? While some, like McNamara, a strong gun control advocate, may simply be motivated by political considerations, there is no evidence that this is true for all officers. Instead, police advice may well logically follow from the resistance experiences of victims with whom officers have had contact. The problem with relying on this sample of resistance cases is that it is substantially unrepresentative of the experiences of crime victims in general—the cases McNamara and other police officers have seen are not like those they have not seen, and the latter outnumber the former by a wide margin.

Most crimes are not reported to the police, and the crimes most likely to go unreported are the ones that involve neither injury nor property loss, i.e., those that had successful outcomes from the victim’s viewpoint. For example, among robberies reported to the National Crime Victimization Surveys, only 24% of those with no injury or property loss were reported to police, whereas 72% of those with both were reported. Likewise, assaults without injury are less likely to be reported than those with injury. By definition, all incidents involving successful defensive gun uses fall within the no-injury, no-property-loss category, and thus are largely invisible to the police.

Consequently, police never hear about the bulk of successful defensive gun uses, instead hearing mostly about an unrepresentative minority of them containing a disproportionately large number of failures. Further, even when they do receive a report of a crime that in fact involved a gun-wielding victim, the victim has strong legal reasons for leaving their own gun use out of their account of the crime. Since most defensive gun uses occur away from the victim’s home, and few victims have the required permits for carrying concealed weapons in public, most gun uses probably involved a crime on the part of the victim.

To conclude that armed resistance is ineffective or dangerous, based on the experiences of this sort of unrepresentative sample of victims, can be called, in honor of Chief McNamara, “the Police Chief’s Fallacy.”

At present, advising victims to not use guns to resist criminal attempts seems imprudent at best, dangerous at worst.

Another variant of this fallacy concerns the frequency of defensive gun uses rather than their effectiveness. It is the belief that because defensive gun uses are rarely recorded in police reports of crimes, they rarely occur. This is fallacious not only because many incidents involving a defensive gun use would not be reported to police at all, but also because even when the crime is reported, victims are unlikely to mention their own legally controversial actions or their possession of legally questionable guns.

When gun-control advocates and public health scholars consider whether keeping a gun for defensive purposes is sensible, they frequently bring up one variant or another of the most nonsensical statistic in the gun control debate.

In 1975 four physicians published an article based on data derived from medical examiner files in Cuyahoga (Cleveland) County, Ohio. They noted that during the period 1958-1973, there were 148 fatal gun accidents (78% of them in the home) and 23 “burglars, robbers or intruders who were not relatives or acquaintances” killed by people using guns to defend their homes. They stated that there were six times as many home fatal gun accidents as burglars killed. (This appears to have been a miscomputation—the authors counted all 148 accidental deaths in the numerator, instead of just the 115 occurring in the home. Although the value of the number does not matter much, the correct ratio was five rather than six.)

On the basis of these facts alone, the authors concluded that “guns in the home are more dangerous than useful to the homeowner and his family who keep them to protect their persons and property” and that “the possession of firearms by civilians appears to be a dangerous and ineffective means of self-protection.”

Eleven years later, Arthur Kellermann and his colleagues unwittingly replicated the Rushforth findings, finding that “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.” The authors then concluded, just as Rushforth et al. did, that “the advisability of keeping firearms in the home for protection must be questioned.”

While conceding that they had made no effort to count “cases in which burglars or intruders are wounded or frightened away,” the authors never acknowledged a far more pertinent and serious omission: lives saved by defensive gun use.

The basic problem that makes these ratios nonsensical is that they are presented as risk-benefit ratios, but in fact do not reflect any benefits of keeping guns for self-protection. If one sets out to assess only the costs of a behavior, but none of its benefits, the results of such an “analysis” are a foregone conclusion. What is so deceptive about the ratio is the hint that killing burglars or intruders is somehow a “benefit” to the householder. This is both morally offensive and factually inaccurate. Being forced to kill another human being, criminal or not, is a nightmare to be suffered through for years. Even police officers who take a life in the course of their duties commonly suffer the symptoms of post-traumatic stress syndrome.

Gun owners do not keep guns for the sake of having a chance to “bag a burglar.” Instead, the benefit of defensive gun ownership that would be parallel to innocent lives lost to guns would be innocent lives saved by defensive use of guns.

As previously noted, less than one in a thousand defensive gun uses involves a criminal being killed. Few purportedly life-saving defensive uses of guns involve killing the criminal, and, conversely, killings of criminals do not necessarily involve saving the life of a victim. Therefore the number of criminals killed does not in any way even approximately index the number of lives saved. It is, however, impossible to directly count lives saved, i.e. deaths that did not occur, so it will never be possible to form a meaningful ratio of genuinely comparable quantities.

This implied cost-benefit ratio is so meaningless that it can fairly be dubbed the “Nonsense Ratio.”

To briefly summarize:

Defensive gun uses by crime victims are three to four times more common than crimes committed with guns;

Victim gun use is associated with lower rates of assault or robbery victim injury and lower rates of robbery completion than any other defensive action or doing nothing to resist;

Serious predatory criminals perceive a risk from victim gun use that is roughly comparable to that of criminal-justice-system actions, and this perception may influence their criminal behavior in socially desirable ways.

A deterrent effect of widespread gun ownership and defensive use has not been conclusively established, any more than it has been for activities of the legal system. Given the nature of deterrent effects, it may never be convincingly established. Nevertheless, available evidence is consistent with the hypothesis that civilian ownership and defensive use of guns deters violent crime and reduces burglar-linked injuries.

Economic injustice, a history of racism, and other factors have created dangerous conditions in many places in America. Police cannot realistically be expected to provide personal protection for every American, and indeed are not even legally obliged to do so. Although gun ownership is no more an all-situations, magical source of protection than the police, it can be a useful source of safety in addition to police protection, burglary alarms, guard dogs, and all the other resources people exploit to improve their security. These sources are not substitutes for one another. Rather, they are complements, each useful in different situations. Possession of a gun gives its owner an additional option for dealing with danger. If other sources of security are adequate, the gun does not have to be used; but where other sources fail, it can preserve bodily safety and property in at least some situations.

People sympathetic toward gun control yet skeptical about its likely impact sometimes note that although a world in which there were no guns would be desirable, it is also unachievable. The evidence raises a more radical possibility—that a world in which no one had guns would actually be less safe than a hypothetical one in which nonaggressors had guns and aggressors did not.

If gun possession among prospective victims tends to reduce violence, then reducing such gun possession is not, in and of itself, a social good. To disarm noncriminals in the hope that this might indirectly help reduce access to guns among criminals is a very high-stakes gamble, and the risks will not be reduced by pretending that crime victims rarely use guns for self-defense.