By Daniel D. Polsby and Dennis Brennen
October 30, 1995
Click here to see the executive summary.
(Online versions of Heartland Policy Studies are are stripped of citations and graphics. If you would like to receive a complete version by snail mail, please e-mail us. Be advised that printed versions are $9, pre-payable by check or credit card.)
In the twenty-year period between 1973 and 1992, the number of privately owned firearms in the United States increased 73 percent--from 122 million to nearly 222 million. Handgun ownership increased by 110 percent, from 37 million to 78 million. In 1994, the latest year for which data are available, there were 231 million firearms in private hands in the U.S., 82 million of them handguns.
Advocates of gun control see in these statistics not the wide distribution of the tools for self-defense or recreation, but rather an enormous potential for private violence. Indeed, if gun owners were substantially more violent than their fellow citizens, the streets would run red with blood. Or more precisely, country lanes would run red with blood, since gun ownership is considerably more prevalent in rural than in urban areas. The fact that rates of violent crime are much higher in urban areas than in rural and suburban areas is just one of many pieces of evidence suggesting that the simplistic, but popular, theory that "guns cause crime" is wrong.
Gun owners are not, in fact, more violent than citizens who do not own guns. Indeed, the evidence suggests that they tend to be a bit less prone to violence. Gun owners are better-educated, more middle-class, more affluent, and somewhat older than average Americans. Gun owners have roughly the same personality characteristics as non-owners, and indeed it would be strange if they did not. As psychologists Edward Deiner and Kenneth Kerber put it:
Since about one-half of the households in the U.S. contain a gun, it seems somewhat unrealistic to attribute severe abnormal characteristics to the average gun owner (unless one is willing to see considerable pathology in most people).
The prevalence of firearms in private hands constitutes a practical limitation on what gun control laws can be expected to accomplish. Some of the more phobic anti-gun activists have urged that all guns should be "outlawed"; but such proposals are no more sensible than asking that toasters or automobiles or pet dogs be "outlawed." At most, one could hope for regulation of firearms ownership and rules for their use, but the feasibility of strict mandates, let alone the proscription of whole classes of commonplace household items, is dubious indeed.
Nevertheless, we do not lack for laws. It is commonly estimated that more than 20,000 federal, state, and local firearms laws are on the books in this country, and more are being written all the time. There is little evidence that these laws have restrained crime. Nor is there much reason to believe that still more laws would produce additional public benefit. The reasons why are examined in this study.
Part 1 offers an overview of several pivotal issues in the gun control debate. First, we briefly address ten popular misunderstandings about crime, guns, and gun control. Then we consider three of these misunderstandings in further depth, concluding:
The problem of "friends killing friends" is much less troublesome than the media and other gun control advocates would have us believe. The evidence suggests that this problem is more accurately described as criminals killing criminal associates.
Mandatory waiting periods (such as those required under the Brady law) affect legitimate users of firearms more than criminals and the mentally ill, against whom waiting periods are supposedly directed.
Drive-by shootings are already against the law; in fact, they are potential death penalty cases under the Illinois criminal code. There is no reason to believe that additional gun control laws aimed at drive-by shootings will provide any increment of deterrence.
In Part 2, The Record on Gun Control, we ask: What has it achieved for Americans to date? Many respected researchers-- James Wright and Peter Rossi, Gary Kleck, Brandon Centerwall, and others--have found little or no direct connection between gun control and crime. A case study of Chicago's experience with gun control closes this section.
Part 3, The Economics of Gun Control, helps explain why the record on gun control is so poor. Gun control laws intend to affect both the supply and the demand for guns in the illegal market by raising the price of obtaining guns in the legal market. Economics--the science of supply and demand--offers an ideal framework for understanding the likely outcome of gun control policies.
We offer a summary and concluding remarks in Part 4.
Gun control is an issue surrounded by (some would say submerged in) myth and misunderstanding. We present here ten myths that are most frequently raised . . . and, from our perspective, most commonly misunderstood.
Myth No. 1: Guns cause crime. A review of the academic literature shows that there is no relationship between the number of guns and the amount of crime in the United States. Criminologists Gary Kleck and E. Britt Patterson reported in 1993 their finding that gun ownership had no significant effect on the rates of murder, assault, robbery, or rape in the U.S. Between 1973 and 1992, the rate of gun ownership in the U.S. increased by 45 percent (from 610 guns per 1,000 people to 887). The homicide rate during that period fell by nearly 10 percent (from 9.4 homicides per 100,000 people to 8.5).
Myth No. 2: Gun control laws reduce crime. Firearms have been regulated with increasing stringency in the United States for most of the past thirty years. Nevertheless, the number of firearms in private hands has increased continuously by many millions per year; handguns have become an increasing proportion of privately owned firearms; and rates of crime, violent crime, and homicide have shown no relationship to the passage or enforcement of gun laws. In their 1993 research, Kleck and Patterson analyze the impact of 19 gun control measures on six categories of violence. In ninety of the resulting 102 relationships, they found no significant correlation between gun laws and violence.
Myth No. 3: Gun control laws stop friends from killing friends. Most murderers and most victims of homicide have criminal records. They are likely to have other criminals as friends and acquaintances. So while it is true that in many cases of homicide the offender and victim are known to each other, it is not true that these "friends killing friends" are the plain ordinary folks often portrayed in anti-gun propaganda. "It is not a slander on the few truly innocent and highly sensationalized victims," writes Dr. Edgar A. Suter and his colleagues, "to note that the overwhelming predominance of homicide victims' are as predatory and socially aberrant as the perpetrators of homicide." Indeed, according to City of Chicago data, the largest and fastest-growing category of relationship between killer and victim is "non-relative, non-friend acquaintance."
Myth No. 4: Gun control laws keep criminals from obtaining guns. In surveys of prisoners, a majority report that they had owned a handgun prior to their imprisonment. But only 7 percent of criminals' handguns are obtained from legitimate retail sources. Three-fourths of felons surveyed report they would have no trouble obtaining a gun when they were released, despite legal prohibitions against firearms ownership by convicted felons.
Myth No. 5: Required waiting periods would prevent some of the most vicious crimes. The Brady waiting period law imposes waiting periods on handguns--the least-deadly type of firearm--while imposing no such restriction on much more deadly, substitutable weapons such as rifles or shotguns. While handguns are preferred by criminals because of their portability and concealability, not every criminal who planned to use a handgun will abandon his criminal plans when confronted by a waiting period. Indeed, for reasons discussed in more detail below (see "Why Waiting Periods Fail"), it is entirely possible that waiting period laws could increase the number of both killings and nondeadly woundings.
Myth No. 6: Guns don't work as self-protection against criminals. In fact, guns are about as valuable to civilians as they are to police officers, and for the same reason. According to criminologists Gary Kleck and Marc Gertz, every year adults use guns for protective purposes 2.5 million times. As many as 65 lives are protected by guns for every life lost to a gun. Each year, potential victims kill between 2,000 and 3,000 criminals; they wound an additional 9,000 to 17,000.
Moreover, mishaps are rare. Private citizens mistakenly kill innocent people only thirty times a year, compared with about 330 mistaken killings by police. Criminals succeed in taking a gun away from an armed victim less than 1 percent of the time. The real utility of defensive firearms, moreover, must surely be far greater, and would be measured not by how many people were shot or even how often a gun was fired, but rather by the deterrent effects of a civilian being armed.
Myth No. 7: Guns aren't needed as self-protection. About 83 percent of the population will be victims of violent crime at some point in their lives, and in any given year serious crime touches 25 percent of all households. The odds are not likely to improve; there is only one police officer on patrol for every 3,300 people. And the courts repeatedly have ruled that government has at most a limited duty to protect individual citizens from crime.
An illustrative case is Warren v. District of Columbia, in which three rape victims sued the city under the following facts: Two of the victims were upstairs when they heard the other being attacked by men who had broken in downstairs. From an upstairs telephone, the two roommates made several calls to the police. Half an hour passed and their roommate's screams ceased; they assumed the police must have arrived. In fact, however, their calls had been lost in the shuffle while the roommate was being beaten into silent acquiescence. When her roommates went downstairs to see to her, as the court's opinion describes it, "For the next fourteen hours the women were held captive, raped, robbed, beaten, forced to commit sexual acts upon each other, and made to submit to the sexual demands" of their attackers.
Having set out these facts, the District of Columbia's highest court nevertheless exonerated the District and its police, noting that it is
a fundamental principle of American law that a government and its agents are under no general duty to provide public services, such as police protection, to any individual citizen.
Myth No. 8: Gun control laws are especially needed to prevent the purchase of Saturday Night Specials and "assault weapons." Inexpensive handguns are involved in only 1 to 3 percent of violent crimes; criminals generally prefer larger caliber and more expensive handguns. Moreover, in the past fifty years no civilian has ever used a legally owned machine gun in a violent crime. And despite their repeated use by drug dealers on television and movies, no Uzi has ever been used to kill a police officer in the United States. Even some gun control advocates concede that so-called assault weapons play a minor role in violent crime. In 1991, 1992, and 1993 combined, there were more than 2,500 criminal homicides in the City of Chicago--only three of which were perpetrated with a true, military-style, "assault weapon."
Myth No. 9: Gun control laws are especially needed to prevent gun accidents in the home. "Gun-control advocates have sought to create the impression that firearm accidents involving children are a large and growing problem," writes the Independence Institute's David Kopel. "Many people mistakenly conclude that children die frequently in gun accidents and that sharp restrictions on gun ownership are necessary to address the problem." In fact, however, the number of gun accidents involving both children and adults has fallen dramatically.
In 1970, 2,406 Americans died from firearms accidents. By 1991, that number had fallen to 1,441--even as the number of guns increased dramatically. Between 1970 and 1991, the annual rate of fatal gun accidents was cut in half, from 1.2 to 0.6 per 100,000 Americans. The death rate from firearms accidents is lower than that from accidental drowning (1.6 per 100,000 in 1991), inhalation and ingestion of foreign objects (1.3), and complications from medical procedures (1.0).
Myth No. 10: Gun ownership is not a constitutional right. The Second Amendment reflects the founders' belief that an armed citizenry (called the general militia ) was a necessary precaution against tyranny by our own government and its army. The idea that government has a constitutional right to disarm the general citizenry is totally foreign to the intent of the Constitutional framers. Samuel Adams, for example, expressed in the Massachusetts convention his intention that "the said Constitution be never construed . . . to prevent the people of the United States who are peaceable citizens, from keeping their own arms." David Kopel summarizes the legal scholarship on this issue:
In the field of legal scholarship, the primary question has been answered: the Second Amendment was plainly intended to guarantee a right of individuals to possess arms. The essential purpose of this guarantee was not to protect sporting uses of guns, but to facilitate resistance to criminal governments, which were seen as simply a larger case of resistance to individual criminals.
Joseph D. McNamara, one-time chief of the San Jose, California, police department, and a member of the board of directors of Handgun Control Inc., has written that "a gun can do strange things to people." He offers this grim picture:
Take a successful businessman who ordinarily has complete control of his emotions. If you stick a gun in his hand when he is under stress, the feeling of power can become overwhelming. This man becomes fearless. He is a different person, and his otherwise good judgment may suddenly desert him. At any moment, this killer personality can present a danger to his wife and loved ones.
Mr. McNamara does not document this extraordinary story, and the present authors would be grateful to receive any confirmation that there may be anything other than a very occasional incidence of unremarkable Jekylls magically transforming into bug-eyed Hydes because of a gun. The evidence, as it now exists, is very much the other way. People who commit murders with guns usually have a police record, as do their victims.
The evidence usually given in support of the McNamara hypothesis is that homicide offenders often know or are related to their victims. That is true, but it hardly proves that these offenders are successful people ordinarily in control of their emotions. Much more plausibly, it shows that members of the criminal underworld often know one another, and that people with predatory, impulsive patterns of behavior attack relatives and associates as well as strangers. The often-repeated proposition that murder usually occurs between friends and family members is bogus--as Adam Walinsky has recently called it, "a twenty-year fraud" perpetrated by statistical distortions and misrepresentations.
In a recent, fairly typical year (1993), fewer than 3 percent of Chicago homicides involved victims in a marital relationship with the offenders, and only 3 percent involved blood relatives. On the other hand, in 25 percent of cases the victim and offender were described by police as "acquaintances" but not "friends," and there were a similar number of cases where no previous relationship at all between victim and offender could be established.
Those data surely understate the criminal histories of both offenders and victims. Only about two-thirds of homicides are ever "cleared" (that is, result in someone being arrested for the crime). Homicides resulting from domestic quarrels are cleared at a much higher rate than killings between gang members or strangers. Furthermore, an unknown proportion of both offenders and victims who are shown as having no police record do in fact have such a record; they only appear not to because of statutes that expunge the records of juvenile offenders.
In other words, it is not true that homicides are commonly committed by ordinary citizens who just fly off the handle. Very seldom are successful people who ordinarily have "complete control" of their emotions implicated in such crimes (or, for that matter, in any violent crimes). Most homicides are perpetrated by and upon socially marginal individuals, many of whom, because of prior run-ins with the law, are already forbidden by law to possess firearms. Controlling the behavior of this cohort through additional gun control legislation is nearly hopeless.
When the Brady law was wending its way toward passage last year, much was said about the potential of waiting periods to restrict the flow of firearms into the hands of persons who should not possess them. It was theorized that the burden on legitimate users would be small, while the inconvenience to criminals and lunatics should keep at least a few of them from impulsively buying a gun. Since the law was passed, various estimates of its effectiveness have been circulated, all premised on the dubious idea that each sale denied represented a life saved or a crime averted.
The reality is undoubtedly quite different, though admittedly difficult to measure. For one thing, the burden on legitimate users is not necessarily de minimis if the need for a firearm develops suddenly, as happened, for example, during the Los Angeles riot several years ago. Civil disturbances of that kind are rare, but when they occur, those caught in the crossfire may very well need immediate access to a firearm for self-defense or to defend family members. A delay of days, or even hours, can and has resulted in otherwise avoidable deaths.
But aside from being burdensome, waiting period laws are apt to be ineffective. People who want guns for illegal purposes are unlikely to be deterred by such a low hurdle as a five-day waiting period, or indeed a waiting period of any other length. As Nobel laureate economist Gary Becker has noted,
[Waiting periods] do little to keep guns out of the hands of teenagers and criminals who obtain their weapons underground, where guns are sold to anyone who can pay for them. This is the route by which arsenals of weapons have found their way into inner cities and elsewhere in the U.S. Guns continue to be smuggled into the illegal market from abroad, from military stock, and from crooked gun dealers.
Confronted with a waiting period, a prospective firearms abuser has two options: either abandon his plans in frustration or find a substitute weapon. Substitutes include obtaining a gun illegally--buying one from a fence or other illicit dealer or simply stealing it outright; obtaining it legally by "borrowing" it from a relative or associate; obtaining an unregulated weapon, such as a knife or club; or, in the outrageously counterproductive case of the Brady law, simply buying a rifle or shotgun instead of a handgun.
No theory predicts how such substitution behavior actually affects rates of violence or the number of resulting crimes. But it is difficult to see how a waiting period law could make matters better, and easy to see how it might make things worse.
Substituting knives for guns. When confronted by someone with a gun, most people, victims and criminals alike, usually surrender, and for this reason they are less likely to be injured than when confronted by someone who is either unarmed or armed with a weapon other than a gun. If waiting period laws cause criminals to substitute lesser weapons for guns, one could expect to see more incidents in which woundings occurred.
Substituting shoulder arms for handguns. Two-thirds of the firearms in private hands are rifles or shotguns, which shot-for-shot are two to three times deadlier than handguns. (Handguns nevertheless are preferred by criminals because of their portability and concealability.) The prospect of appreciable numbers of psychopaths buying, "borrowing," or stealing shoulder weapons in preference to handguns should be deeply troubling to any policymaker. Yet it is almost certainly to be expected from the Brady law, which imposes waiting periods on the least-deadly firearms while imposing no such restriction on much more deadly substitutes.
Summary. It is entirely possible that waiting period laws could result in increased numbers of killings and non-lethal woundings. The former should increase because a larger proportion of criminals might be armed with rifles or shotguns, and the latter should increase because more criminals might be armed with knives or weapons other than a gun.
Because of their seemingly random nature and increasing frequency, drive-by shootings are among the most disturbing of crimes. In many cases these begin as affairs of honor, part of the tit-for-tat of the criminal life. They also may have an economic motive, as when rival gangs try to perfect a monopoly over drug sales or other illegal activity in a particular geographic area. Innocent bystanders are often injured or killed, and gun control advocacy groups seldom miss an opportunity to take maximum propaganda advantage of incidents of this kind.
It can hardly be disputed that "something should be done." But attempts to reduce drive-by shootings by restricting access to firearms are doomed to failure. It must be borne in mind that in all cases of drive-by shooting, the weapons themselves and the use to which they are being put are already illegal and carry heavy penalties.
Under Illinois law it is a felony to carry a loaded handgun in a vehicle. If the offenders are under 18, mere possession of a handgun is a Class A misdemeanor, punishable by a sentence of up to one year in prison. If the offenders are under 21 and convicted misdemeanants, or adjudged delinquent, their possession of any firearm or ammunition is a violation, punishable by a sentence of up to one year. If the offenders are convicted felons, possession of any firearm is a Class 3 felony, punishable by a sentence of two to five years.
If the drive-by shooters are carrying an automatic weapon, they are in violation of federal law and face a penalty of up to ten years imprisonment. Possession is also illegal under Illinois law. (It is more likely, however, that the offenders were carrying semi-automatic weapons, often mistakenly referred to as "assault weapons" although mechanically no different from many popular hunting and target rifles.)
The prospect of all these penalties appears not to deter drive-by shooters, and why should it? They are, after all, on their way to commit first-degree murder, punishable by no less than a death penalty. Further gun control laws could hardly be expected to offer more deterrence than that.
What is the evidence that gun control laws affect rates of violent crime? Scholars seem to be deeply and often bitterly divided on this question. Whether one thinks there is abundant evidence of efficacy or none at all depends on how one evaluates the many studies that purport to show this effect.
The basic questions here are empirical, but--as is commonly the case with social science research, where data can be gathered but experiments cannot be conducted--the facts are difficult to ascertain and have become deeply embroiled in controversy. In any event, facts must first of all be organized by a theory.
The theory favoring gun control begins with two plausible and related observations: First, that firearms facilitate the infliction of death or great bodily harm because they inflict more dangerous wounds than weapons that might be substituted for them; and second, that many killings or grave woundings are not the product of a literal intention to inflict death or great bodily harm, but are merely the byproduct of the fact that a gun was handy and was used. Colin Loftin and his colleagues have called this latter proposition the "Zimring-Cook" effect, referring to the work of researchers Frank Zimring and Philip Cook.
The opposite theory, attributable primarily to Florida criminologist Gary Kleck, stresses that there are two sides to the strategic ledger. Guns do inflict dangerous wounds, but they are also useful for self-defense. Moreover, while firearms are, wound-for-wound, more dangerous than other kinds of weapons, hostile confrontations with firearms are much less likely to result in woundings in the first place than face-offs without arms or with lesser weapons. Because firearms have the potential to reduce the variance between antagonists' capacity to do one another harm, they might well contribute to a world in which there is less predatory behavior rather than more.
Many people believe that the complete disappearance of firearms would make for a more peaceable world. Still, there is no reason to believe that there would be a simple, monotonic relationship between diminishing firearms and diminishing violence. For example, if every year we could eliminate 10 percent of the nation's existing stock of firearms, and in Year One, the 10 percent that we eliminated were those carried by police officers, few people would expect to see a 10 percent reduction, or indeed any reduction at all, in violence. Of course no one proposes disarming the police first. But the point of the illustration is more general: Potential victims and their protectors, as well as aggressors, are disarmed by gun control laws. It is not the sheer number of firearms, but how and to whom they are distributed, that determines whether there is more violence and crime or less.
Much of the "evidence" on which both pro- and anti- gun control partisans rely is simplistic. It is noted, for example, that England has fewer guns and much less crime than the U.S. . . . but Switzerland has guns in almost every home, and much less crime than the U.S. Moreover, there are more guns per person in the American South than in the North, and more homicides . . . but there are more guns per person in rural areas than in urban areas, and fewer homicides.
The thoughtful person immediately recognizes that England, Switzerland, and the United States, as well as urban and rural areas within the United States, differ in many ways other than in just the prevalence of guns. There are differences in culture, history, and ethnic mix, to name only a few. More variables must be taken into account.
One method social scientists use to account for the effects of many variables is "multiple regression analysis." It is a statistical method that attempts to disentangle the effects of each of many variables that are thought to affect the phenomenon to be explained. If a variable seems to be explanatorily useful, it is termed "statistically significant."
In the most comprehensive and careful such analysis to date, criminologists Gary Kleck and E. Britt Patterson reported in 1993 their findings on the relationship between guns, violence, and gun control. Kleck and Patterson sought to avoid the many methodological weaknesses that had in the past plagued studies of the effect of gun control laws on violent crime rates. Their analysis "covered all forms of violence which involves guns, encompassed every large (over 100,000 population) city in the nation, and assessed all major forms of existing gun control in the U.S."
The preponderance of statistical evidence contradicts the notion that more gun control laws will reduce crime. If they could, we would have a simple solution to the crime problem. But they do not, and thus we do not. Believing that gun control will reduce crime is wishful thinking.
There is no shortage of wishful thinking among gun control advocates and the nation's popular media. Nearly twenty years ago, the National Institute of Justice commissioned a three-year study to take stock of what was then known about firearms, crime, and violence in the United States. The authors of that study neatly summarized the popular case against privately owned firearms. The indictment, they wrote,
includes the following particulars: (1) Guns are involved in an astonishing number of crimes in this country. (2) In other countries with stricter firearms laws and fewer guns in private hands, gun crime is rare. (3) Most of the firearms involved in crime are cheap Saturday Night Specials, for which no legitimate use or need exists. (4) Many families acquire a gun because they feel the need to protect themselves; eventually they end up shooting one another. (5) If there were fewer guns around, there would obviously be less crime. (6) Most of the public also believes this and has favored stricter gun control laws for as long as anyone has asked the question. (7) Only the gun lobby prevents us from embarking on the road to a safer and more civilized society.
"Both the senior authors of this study"--Professors James Wright and Peter Rossi--"once shared" this point of view, they wrote; but "the more deeply we have explored the empirical implications of this indictment, the less plausible it has become." As of the early 1980s, the evidence in support of the indictment, they found, was unimpressive and shot through with contradictions. In the decade since, the indictment has changed not at all, but a fresh crop of research has been directed at proving the case.
Many works purport to link the wide distribution of private firearms to high rates of homicide and suicide, and the presence of gun control measures to reductions in the rates of such violence. Our discussion here will concentrate on five widely cited papers, published in The New England Journal of Medicine, which have become the standard authorities for the proposition that guns hurt and gun control helps.
First are two articles by John Sloan and others, which purport to explain the different homicide and suicide experiences of Seattle, Washington and Vancouver, British Columbia, by the differences in handgun regulations between the two cities. Second, in 1986, Arthur Kellermann and Donald Reay concluded that a firearm kept in the home was 43 times more likely to kill a family member than a criminal. In October 1993, Kellermann and several collaborators, using new data, concluded that a person increases his or her chances of becoming a homicide victim by keeping a gun at home. Finally, a paper by Colin Loftin and collaborators shows the decrease of both homicide and suicide following enactment of the District of Columbia's restrictive 1976 gun control law, which forbade the purchase, sale, transfer, or possession of handguns by civilians.
The news media have accepted these studies as firmly establishing the scientific truth that widespread possession of firearms leads to high rates of murder and suicide, and that restrictive gun control laws will lead to lower rates of murder and suicide. Closer inspection, however, reveals that these studies are plagued by methodological and interpretive weaknesses, and do not support the propositions for which they are so often cited.
The implied premise of the Seattle-Vancouver studies is that the two cities are matched like identical twins, differing from one another along only two dimensions: their homicide-suicide rates and how they regulate handguns. If this were the case, one might infer some sort of connection between the two variables. But there are numerous objections to proceeding in this way.
In the first place, it is not possible using the methodology of these studies to tell what causes what. Does possession of a firearm pose a risk factor for suicide, or is it rather that people bent on self-destruction prefer to obtain a firearm a cheap and relatively certain means of carrying out their intentions?
Secondly, it is erroneous to infer that Seattle and Vancouver are a "matched set." Substantial demographic differences between the two cities exist. Sloan and his colleagues did not include a Mantel-Haenszel analysis (that is, a standard test of statistical significance) in order to ascertain whether those differences could account for differences in their observed homicide and suicide rates. Nor have the study's authors made their data available to permit others to perform such an analysis.
Moreover, Sloan and his colleagues did not attempt to ascertain whether the prevalence of handgun ownership actually differed, and if so how extensively, among the two populations. Handguns have been restrictively regulated in Canada only since 1977, less than ten years before the data for the studies were gathered. Yet handguns have useful lives of a hundred years or more. It is not known whether (or by how much) Vancouver and Seattle actually differ with respect to the ownership of handguns. In fact, one student of the problem has suggested that the prevalence of such weapons may actually have increased in Canada. In any case, one certainly cannot reason that, simply because the laws are different on either side of the border, therefore the actual behavior of the people is different.
Finally, even if the two cities were literally identical twins, with only handgun regulation as a variable, it would make little sense to allow any conclusion to turn on the results of a single pair-wise comparison. To illustrate: Suppose a researcher followed a pair of identical twins through their lives and determined that the only difference between them was in politics (one was a Socialist, the other not). Suppose further that one of them died at age 82, while his brother lived to age 85. No one would propose politics as a risk factor for mortality on the basis of any such observation--certainly not if it were possible to evaluate the variable on the basis of other readily available data.
For comparisons between the United States and Canada, such data are readily available. Thus, Brandon Centerwall was able to compare the homicide rates of urban and rural populations in Canadian provinces and in each adjacent American state. He found no consistent differences between them: "When Canadian provinces and adjoining U.S. states are compared, three- to tenfold differences in the prevalence of handguns have not resulted in consistently different rates of criminal homicide."
Studies of this kind do not lend themselves to eye-catching headlines and oversimplified certitudes about the causes of violence and means to prevent it. Yet methodologically, this sort of study is superior to the Sloan et al. sort, because it includes all available data, thus avoiding the question of why a particular pair of cities was selected for comparison. Robert Mundt's study of Canadian homicide and suicide rates before and after the enactment of the 1977 law is basically confirmatory: "When compared with the United States, trends in Canada over the past ten years in various types of violent crime, suicide, and accidental death show no dramatic results and few suggestions of perceptible effects of the 1977 Canadian gun control legislation."
In October 1993, the New England Journal of Medicine published an article by Arthur Kellermann et al. which was widely received as the definitive statement of the proposition that firearms kept in the home not only fail to increase the security of the household, but actually are a risk factor for becoming a victim of homicide. This 1993 paper followed an earlier (1986) effort by Kellermann and Donald Reay, also published in the New England Journal of Medicine, which is the source of the most famous statistic in the gun control debate: that a firearm kept in the home is 43 times more likely to be used to kill a member of the household than to kill a criminal intruder.
Evaluation of the 1986 study. There are at least two important difficulties with the 1986 study and the "43 times" statistic to which it gave rise. The first, often-overlooked problem is that more than 85 percent of the Kellermann-Reay gunshot victims were suicides and not victims of domestic quarrels. The instrumentality-dependency of suicide is a hotly debated topic, but at present the evidence is very thin for the hypothesis that any appreciable portion of observed suicides is a function of the availability of firearms.
It is always possible, but remains entirely conjectural, that a handy firearm does raise the rate of suicide. If so, this would be an example of the "Zimring-Cook" effect--the proposition that a certain percentage of fatal incidents begin ambiguously so far as the intentions of the actor are concerned, and result in death only because an almost-always deadly weapon like a gun is ready at hand.
It may well be true that a Zimring-Cook effect occurs in some contexts, but it is difficult to see why suicide would be one of them. Internationally, rates of suicide appear to be quite independent both of gun control laws and patterns of firearms ownership in civilian populations. Very high rates of suicide are found, for example, in firearms-free populations like Japan--but of course the rates of suicide in such places might be even higher if firearms were more readily available. And low rates of suicide are found in such relatively well-armed jurisdictions as Israel and New Zealand--but of course those rates might be still lower were firearms absent. Such prospects are impossible to rule out with existing research techniques.
If, as a concession to plausibility, one eliminates the suicide data from the 1986 Kellermann-Reay study, the "43 times" statistic is transfigured into one less imposing: a firearm kept in the home is six times more likely to be used to kill a member of the household than to kill a criminal intruder. But even this number is of precarious significance, for it embodies the dubious assumption that comparing "body counts" is a meaningful way to report the usefulness of firearms.
Comparative body counts are not ordinarily used to measure the utility of firearms or the use of force. One does not measure the effectiveness of a police department, for example, by comparing the number of officers and criminals killed over some time period. Rather, one asks what effect the police have had on the rate of crime. Similarly, one should not compare the number of burglars or other intruders and civilians killed by domestic firearms, but rather ask how many burglars or other intruders were driven away or deterred by the firearm.
Any meaningful tally of firearms "use" must include not merely the fraction of cases in which someone was killed, but also cases in which there was a wounding and, for that matter, the probably much larger number of cases in which a weapon was used simply to threaten, but was not discharged at all. Indeed, the case is broader even than that. We should also try to estimate the number of crimes that did not occur at all because of the prospect of meeting armed resistance. Research by criminologist Don Kates offers reason to believe that the probability of encountering an armed defender does enter into a potential criminal's calculations.
Evaluation of the 1993 study. The 1993 study by Kellermann et al. raises a number of different problems. The "case-control" methodology used in the study risks generating spurious correlations if the researcher fails to take into account subgrouping, sometimes called social stratification.
By way of example, the U.S. population can be said to consist of two subgroups. The first subgroup--a relatively small minority of the total population--is composed of career criminals, gang members, and others with a history of criminal activity. This subgroup has a high risk of homicide and a relatively high gun ownership rate. The second subgroup--the vast majority of the total population--is the general law-abiding public. This subgroup has a low risk of homicide and a lower gun ownership rate than the first.
Henry Schaffer, professor of genetics at North Carolina State University, explains:
There is no causal relationship between gun ownership and homicide in either subgroup. . . . However, when we put the two groups together into the single population they compose, . . . an association between gun ownership and homicide [appears]. This is not due to gun ownership having a causal effect, but rather there is a "confounding" variable of subgroup membership, and gun ownership is associated with that subgroup.
In principle, it would be possible to ascertain whether the Kellermann results can be explained by subgroup analysis, but as with the Seattle-Vancouver study, the authors have not made their data available to allow this analysis to be performed. Lawrence Southwick, professor of management at State University of New York (Buffalo), has pointed out that because of the authors' unwillingness to publish their data, it is "not possible to replicate the statistical tests nor to improve on them." Thus the results of the study must be accepted "on faith, an attitude not in keeping with good science."
A second problem with the 1993 study is that its sample is not well-chosen. As Southwick points out,
Any successful use of a gun for self-defense is excluded from the sample. If a potential victim used a gun to threaten or to shoot an intruder, that was not included in the sample. Only a person who was killed in his own home was included. Consequently, the authors' statement that our methodology was capable of demonstrating significant protective effects of gun ownership as readily as any evidence of increased risk' is clearly false; any protective effect was deliberately excluded.
More fundamentally, it appears that, contrary to the authors' claims, not all the homicides meeting the study's "in the home" criterion were in fact included in the study. As Henry Schaffer points out:
There were 444 homicides meeting the "home" criterion. [Twenty-four] were excluded for "various reasons," leaving 94.6 percent. But then 7 percent were dropped because of a failure to interview the proxy, and an additional 1 percent because of a failure to find a control, leaving 388 matched pairs. The authors state, "Although case-control studies offer many advantages over ecologic studies, they are prone to several sources of bias. To minimize selection bias, we included all cases of homicide in the home, and rigorously followed an explicit procedure for randomly selecting neighborhood control subjects." [M]any would be willing to allow 87.4 percent to be described as "all." However, this is not the end--even though there were 388 matched pairs, it appears that the study did not obtain complete data on all of them, and the multivariate analyses used require complete data, and so there were only 316 matched pairs used in the final analyses. This represents 71.2 percent of the homicides. It is very difficult to accept that "all" fairly describes this 71.2 percent.
A third problem with the study is the possibility that the fear of being killed might inspire a person to arm himself. Actually, that possibility did occur to the study's authors, since they included a specific disclaimer about it in the article's text. But Dr. Kellermann, the study's principal author, seems never to mention the disclaimer in any of his many media interviews, and as a consequence the point gets quite lost in the public conversation. That is unfortunate, for rational self-armament by criminals is a cogent explanation of the association between firearms ownership and homicide. Peter Reuter and Mark Kleiman estimated that drug dealers are at great risk of being murdered, far higher than the risk to which ordinary people are exposed. Homicide data from Chicago are basically confirmatory: In recent years 60 percent or more of murder victims have had police records.
Dr. Kellermann has since refused to meet this point, but rather offered an obfuscation. "It is relatively easy," he wrote, "to establish cause and effect when the victim is dead on the floor with a bullet in his head." But of course there was never a question whether a bullet in the head might "cause" someone's death. The question of causation has always been whether being a drug dealer or criminal, as distinct from being simply a gun owner, increased the risk of being victimized by customers, suppliers, or other (criminal) business associates, whether inside the home or out of it. So long as a well-founded fear of death might lead someone both to arm himself and to be at greater risk of being murdered, the correlation upon which Dr. Kellermann places so much weight simply cannot support the inference he seeks to draw.
Data from the District of Columbia are persuasive or unpersuasive about the efficacy of gun control laws, depending upon what time period is chosen for measurement. Colin Loftin and his coauthors tracked the experience in the District of Columbia for seven years after its 1976 gun control law was enacted and concluded:
In the District of Columbia, the mean frequency of both suicides and homicides by firearms declined by about one-quarter in the period after the law went into effect. In contrast, none of the comparison time series showed declines of similar magnitude during the same period. . . . The adjacent areas in Maryland and Virginia, which were not subject to the change in gun regulations, did not have declines in gun-related homicides and suicides similar to those observed in the District of Columbia. The best explanation for the District of Columbia data is the weapons-choice theory developed by Zimring, Cook, and others. According to this view, assaults, whether against others or self-directed, vary with respect to the intent to kill. Some are characterized by a sustained, single-minded determination, whereas in others the intention is episodic and ambivalently motivated. If the resolve is weak or short-lived, the relative frequency with which a particular type of weapon is used will be influenced by its availability. The key element in the theory is that firearms are more likely to cause death than are other weapons that are likely to be substituted. It follows that even if there is no change in the number of assaults or suicide attempts, a reduction in the availability of guns will result in a reduction in the number of deaths.
Loftin et al. professed surprise at the magnitude and suddenness of the drop-off in gun-related bloodletting following the enactment of the District's gun control law. Yet, as is generally well known, only a few years later the District of Columbia is experiencing a much higher murder rate than that recorded before the passage of restrictive gun control legislation. It is a chronic liability of time-series analyses that the results they generate are highly sensitive to the time periods chosen for measurement. Indeed, choosing a longer measurement period would have shown an association between restrictive gun control laws and increases in the homicide rate.
Loftin and his colleagues state that the introduction of new variables, such as turf wars among drug dealers, could explain subsequent increases in homicide rates without necessarily invalidating the possibility that, in the absence of the restrictive law, things might have been worse. Indeed so--but this argumentative escape hatch simply highlights the abiding uncertainty of this sort of social science research. A policy can be manifestly a failure, yet its proponents can proclaim it a great success because of how much worse things would have been--conjecturally--had no such policy been in effect. By such "guesstimative" expedients, evidence that is merely suggestive and preliminary is translated into irrefutable proof.
Yet it remains surprising that the effect that Loftin and his coauthors believed they had observed had occurred so swiftly. The theory behind gun control laws--that they "thin out" the number of weapons in circulation--can hardly be supposed to operate over any short run, let alone instantaneously. Guns do not self-destruct because a law was passed, and only the most conscientious and law-abiding would have disposed of newly outlawed weapons. In fact, as was true of the Seattle-Vancouver studies discussed earlier, Loftin did not say whether and how much the law actually reduced firearms ownership. Such data are unavailable. Yet without them, it is hardly possible to understand and interpret the supposed relationship between the distribution of firearms and their rates of abuse.
The City of Chicago has one of the nation's most restrictive gun control laws. Since April 1982 it has been illegal to purchase or register any handgun within the city. At the time Chicago's law was passed, then-Mayor Jane Byrne touted it as an anti-crime measure. How has her claim fared through the past dozen years?
A review of the Chicago Murder Analyses from 1965 to 1992 provides information on this question. The Murder Analyses are compiled annually by the Chicago Police Department, offering painstaking detail about the number of murders committed in the City of Chicago, the types of weapons used in those murders, the age of offenders and victims, and much more.
Figure 1 shows the number of murders in the City of Chicago between 1965 (seventeen years before the city's gun control law was enacted) and 1992 (ten years after gun control). Between 1965 and 1974 there was a steady increase in the number of murders, with 1972 being the only exception. Between 1974 and 1990, the number of murders stayed within a reasonably narrow range, with a fairly dramatic fall in--1982 the year gun control was passed. Just five years later, in 1987, the number of murders in the city began to climb steadily. Indeed, by gun control's tenth anniversary, the number of murders in the city was back where it had been a decade before gun control.
Figures 2 and 3 narrow the analysis to include only firearms (Figure 2) and handguns (Figure 3). The two figures closely track Figure 1: steady increase until 1974; stabilization until 1981; stabilization again at a lower level for five years; and then a steady increase beginning in 1989.
What all three graphs show is that the number of murders ebbs and flows with little apparent respect for gun control laws. The national gun control act of 1968 appears to have had little effect on murders with firearms in Chicago: The number of murders committed with handguns rose dramatically in the years following its passage. The number of murders with handguns was falling in Chicago before passage of the city's 1982 gun control law. That year, the number of murders fell precipitously. Was this evidence of the gun control law working? If so, upon what theory? Were there suddenly fewer guns in circulation? Were criminals, heedless of the state's murder laws before the city passed its ordinance, more careful once handgun possession became a misdemeanor? Then what?
Then perhaps nothing. After all, the national murder rates rose very rapidly from the mid-1960s through 1979, and then began falling. Through most of the 1980s national murder rates declined, as did the rates in Chicago. Then in the late 1980s, national murder statistics began to trend up. Chicago's numbers did likewise. Currently, after thirteen years under a strict handgun ban, handgun murders and murders of all sorts are at record levels in the city.
In light of this record, calling for more, or yet-more-stringent, gun control laws begins to seem like neurotic behavior.
Some gun control advocates believe that a reduction in the total number of firearms will reduce crime. Their implicit model is:
A reduction in the general availability of firearms could be accomplished in a number of ways, including raising taxes on firearms and/or ammunition, restrictive licensing of firearms owners, waiting periods, stricter reporting requirements for firearms dealers, higher dealer fees, etc. All of these methods operate by making it more difficult and/or more expensive for the general public to acquire firearms. All of these methods directly affect only the legal market for firearms. If enforced, they will indirectly affect the illegal or black market by driving up the price of firearms. Firearms still would be available to criminals. But the price would be higher; one would therefore expect to see potential criminals purchase fewer guns and reserve their firearms for "higher-valued" uses. The key questions are: how many fewer guns would criminals buy; how would they channel their weapons to higher-valued uses; and how would the firearms purchasing and retaining behavior of non-criminals change in response to this price increase?
There are sound theoretical reasons to expect only a small decrease in the flow of guns to criminals. The reasons have to do with the characteristics of the supply and demand for firearms in the illegal market.
How many fewer firearms will be demanded by criminals will depend upon the responsiveness of criminal buyers to a price increase, or what economists call the elasticity of demand. If demand is responsive (elastic), a price increase will result in a relatively large decrease in the quantity demanded. If demand is unresponsive (inelastic), a price increase will result in a relatively small decrease in the quantity demanded.
Elasticity depends upon the availability, cost, and other characteristics of substitutes for the item in the intended use. If there are many good substitutes, demand will tend to be elastic (responsive); if there are few good substitutes, demand will tend to be inelastic (unresponsive).
For use in violent crime--armed robbery, turf wars, and so on there are few good substitutes for firearms. Economists Steven Balkin and John F. McDonald explain:
. . . because handguns are easily concealed and are very effective weapons for obtaining victim compliance, other weapons such as long guns and knives are imperfect substitutes for handguns. . . . While the elasticity of demand for handguns may vary with the type of illegal intent, the aforementioned conjectures lead to the expectation that the overall elasticity of demand for handguns for criminal purpose is relatively low.
In other words, the quantity of firearms sought by criminals will not decrease much in response to a price increase.
Analysis of the demand for firearms reveals a second problem with gun control proposals. Even if one assumes, conservatively and counterintuitively, that criminals and non-criminals begin with an identical desire to obtain a gun, they will not end with an identical demand, because (by definition) criminals plan definitely to use their guns (or in any event to have control over whether and when they will use them), whereas people who want guns for self-defensive purposes plan to use them only contingently--that is, in the unlikely event that an appropriate occasion for using them should present itself. According to Balkin and McDonald, "recreational demand has the relatively highest price elasticity, followed by self-protection demand, and offender demand has the relatively lowest elasticity."
Thus, it appears that gun control laws will have the effect not of disarming criminals, but rather increasing the ratio of firearms-holding criminals to non-criminals. Unilateral disarmament does not have a good reputation in international affairs, because such a tactic on occasion has invited aggression by the better-armed party. What reason do we have to suppose that domestic unilateral disarmament would be more successful? The result of gun control is likely to be more violence against non-criminals, not less.
This is hardly a radical insight. Over 200 years ago, in 1764, Italian criminologist Cesare Beccaria wrote:
The laws that forbid the carrying of arms . . . disarm those only who are neither inclined nor determined to commit crimes. . . . Such laws make things worse for the assaulted and better for the assailants; they serve rather to encourage than to prevent homicides, for an unarmed man may be attacked with greater confidence than an armed man.
The price increase caused by gun control will affect not only the demand side of the market for illegal firearms, but also the supply. As is true of markets for other goods, an increase in the selling price of illegal firearms will cause an increase in the number of firearms supplied to the illegal market. How large an increase depends upon the elasticity of supply.
Unlike demand, the supply of firearms to the illegal market is likely to be quite elastic. Gun control would eliminate or heavily regulate the legal market for guns. But reducing the supply of something in the legal market is likely to increase its supply in the illegal market. Consider the many possible sources of supply for the illegal market:
Any of these sources could be increased substantially, and would be increased if a price rise made it profitable.
With an inelastic demand and elastic supply, we get a market that looks like Figure 4. At a price of P0, the number of firearms demanded is (Q0).
If the supply of firearms were decreased by the enforcement of gun control laws, the supply curve would shift to the left. Figure 5 depicts this new market. At the old price (P0), the quantity of firearms demanded would fall to Q1, if demand were perfectly elastic. But because demand for firearms is inelastic (unresponsive to price increases), prices will rise to P2, and the quantity demanded would fall only slightly to Q2.
If new gun control laws succeeded in reducing the gross quantity of guns going to the illegal market by 10 percent (before allowing for price effects), reasonable assumptions produce an estimate of the net change in the quantity of weapons going to criminals of less than 2 percent. That is shown in Figure 5 as an initial quantity of 100 (Q0) and a final quantity of 98 (Q2).
Furthermore, even this small quantity reduction would come from the illegal purchasers least wedded to their guns: those least likely to commit violent crimes. The more violent criminals will still have weapons. As Wright and Rossi have noted:
Here it may be appropriate to recall the First Law of Economics, a law whose operation has been sharply in evidence in the case of Prohibition, marijuana and other drugs, prostitution, pornography, and a host of other banned activities and substances--namely, that demand creates its own supply. There is no evidence anywhere to show that reducing the availability of firearms in general likewise reduces their availability to persons with criminal intent, or that persons with criminal intent would not be able to arm themselves under any set of general restrictions on firearms.
The conclusion is that changes in the supply of weapons to criminals induced by restrictions on the general public are unlikely to result in any noticeable decrease in criminal armament. Consequently, no noticeable impact on crime will occur.
Between 1973 and 1992, the rate of gun ownership in the U.S. increased by 45 percent--from 610 guns per 1,000 people to 887. If firearms caused crime and violence, one should expect to see gun owners noticeably more criminous and more violent than those who do not own guns. One finds no such thing. In fact, the population as a whole has been growing noticeably less violent.
As criminologists well appreciate, the problem of crime and violence in this country has become largely concentrated in the poorest neighborhoods of large and medium-sized cities. Young men, African Americans in particular, have experienced a startling increase in victimization over the last decade.
None of this comes as much of a surprise if one considers the scissors of (1) poor life chances in the middle-class world (owing to the disintegration of families, poor educational opportunities, and so on) and (2) growing opportunities in the drug trade that result from our ever-more-vigorous efforts to suppress drug use by suppressing supply. The most important reason for criminal behavior is that the income that offenders can earn in the world of crime, as compared with the world of work, all too often makes crime appear to be the better choice.
"We are now reaping the consequences of 30 years spent talking about guns rather than doing something effective about poverty and hopelessness," writes David Kopel. "If we really want to reduce the disease of violence, it is time for us to start thinking about how to strengthen families and foster individual responsibility, and it is time to abandon the unscientific crusade against guns."
Firearms are nowhere near the root of the problem of violence and arguably are almost completely divorced from it. As long as people come in unlike sizes, shapes, ages, and temperament; as long as they diverge in their taste for risk and their willingness and capacity to prey on other people or to defend themselves from predation; and, above all, as long as some people have little or nothing to lose by spending their lives in crime, dispositions to violence will persist--and increasingly strict gun controls will do little if anything to improve matters.
Daniel D. Polsby is Kirkland & Ellis Professor of Law at Northwestern University School of Law in Chicago. Dennis Brennen is professor of economics at Harper College in Palatine, Illinois.
[ Heartland Policy Studies | Home page | E-mail Heartland | Membership information ]