Many of us assume we must either oppose or support gun control. Not so. We have a range of alternatives. Even this way of speaking oversimplifies our choices since there are two distinct scales on which to place alternatives. One scale concerns the degree (if at all) to which guns should be abolished. This scale moves from those who want no abolition (NA) of any guns, through those who want moderate abolition (MA) - to forbid access to some subclasses of guns - to those who want absolute abolition (AA). The second scale concerns the restrictions (if any) on those guns that are available to private citizens. This scale moves from those who want absolute restrictions (AR) through those who want moderate restrictions (MR) to those who want no restrictions (NR) at all. Restrictions vary not only in strength but also in content. We could restrict who owns guns, how they obtain them, where and how they store them, and where and how they can carry them.

Our options are further complicated by the union of these scales. On one extreme no private citizen can own any guns (AA, which is functionally equivalent to AR), while at the other extreme, every private citizen can own any gun, with no restrictions (NA+NR). But once we leave those extremes, which few people hold, the options are defined by a pair of coordinates along these distinct scales. While most people embrace positions on the "same" end of both scales, others embrace more exotic mixtures: some will want few weapons available to private citizens, but virtually no restrictions on those guns that are available (MA+NR), while others may prefer making most guns available, but want to seriously restrict them (NA+MR).

So our choice is not merely to support or oppose gun control, but to decide who can own which guns, under what conditions. Although I cannot pretend to provide a definitive account here, I can isolate the central issues and offer the broad outline of an appropriate solution. To simplify discussion, I adopt the following locutions: those opposed to most abolition and most restrictions advocate a "serious right to bear arms," while those supporting more widespread abolition and more substantial restrictions are "gun control advocates." This simplification, of course, masks significant disagreements among advocates of each position.

Justifying Private Ownership of Guns

A Moral Question

Do citizens have a "serious right to bear arms"? This is a moral question, not a Constitutional one. For even if the Constitution did grant this right, we should determine if there are sufficiently compelling arguments against private gun ownership to warrant changing the Constitution. On the other hand, if this were not a Constitutional right, we should determine if there are strong reasons why the state should not ban or control guns, and if these reasons are sufficiently compelling to make this a constitutional right. Most defenders of private gun ownership claim we do have a moral right - as well as a Constitutional one - and that this right is not an ordinary right, but a fundamental one.

(i) A fundamental right

If they are correct, they would have the justificatory upper hand. Were this a fundamental right, it would not be enough to show that society would benefit from controlling access to guns (Hughes, T. C. & Hunt, L. H. 2000). The arguments for gun control would have to be overwhelming. Yet there is also a hefty cost in claiming that this is a fundamental right: the evidence for the right must meet especially rigorous standards.

What makes a right fundamental? A fundamental right is a non-derivative right protecting a fundamental interest. Not every interest we individually cherish is fundamental. Since most interests are prized by someone, such a notion of "fundamental interest" would be anemic, serving no special justificatory role. Fundamental interests are special: they are integrally related to a person's chance of living a good life, whatever her particular interests, desires, and beliefs happen to be. For example, living in a society that protects speech creates an environment within which each of us can pursue our particular interests, goals, needs, and development, whatever our interests happen to be. Is the purported right to bear arms like this paradigmatic fundamental right?

Even if it were, that would not straightforwardly establish that it is impermissible to abolish or restrict private ownership of guns. After all, fundamental rights standardly have conditions, boundaries, or restrictions on them. Some rights, like the right to vote, are conditional upon reaching a specified age, and they can be forfeited by emigrants and imprisoned felons. Additionally, most right tokens can be restricted or overridden when the exercise of that right harms others. For example, my right to free religious expression gives me wide discretion in how I exercise my religion. I can remove my kids from high school and exclude them from selected school activities (Wisconsin v. Yoder, 406 U.S. 205 [1972]; Moody v. Cronin, 484 F. Supp. 270 [1979]). I can sacrifice animals (Church of the Lukumi Babalu Aye v. City of Hialeah, 508 U.S. 520 [1993]). Nonetheless, it does not permit me to sacrifice humans. Nor does my right to free speech permit me to slander someone or to preach outside her window at 2:00 a.m. Tokens of fundamental rights may be restricted to protect others from serious harms arising from the exercise of those rights.

Of course rights would not be worth much if they were straightforwardly subject to the wishes of the majority. We fiercely defend fundamental right types although their tokens sometimes undercut society's interests. We cannot restrict or put conditions on fundamental rights except for compelling reasons, and individuals cannot forfeit their fundamental rights (if they can forfeit them at all) except for overwhelming reasons. Still, although tokens of a right sometimes run counter to the majority's wishes (Dworkin, R. M. 1977), we should not infer that rights standardly undermine the public interest. Fundamental rights (freedom of speech, freedom of association, etc.) benefit society as well as individuals. Permitting free speech, religion, and association is the best - and arguably the only - way for society to uncover the truth (Mill, J. S. 1978). Of course, not every right has such a significant social payoff -- although most fundamental rights do. Still, we minimally assume fundamental rights (right types) do not harm society.

This provides a framework for evaluating people's claims that a right is fundamental. Advocates must show that and how granting the right protects individuals' fundamental interests, and they must be prepared to respond to objections that granting that right type will harm society. These are serious obstacles for gun advocates. It is difficult to see that a serious right to bear arms satisfies either of these requirements, let alone both.

First, I see no compelling reason to think that owning a gun is a fundamental interest. Other fundamental interests are necessary to one's flourishing no matter what her particular desires, interests, and beliefs. It is difficult to see how this is true of guns. Moreover, the interests protected by paradigmatic fundamental rights - our interests in unfettered speech, freedom of religion, and freedom of association - are not merely means to my flourishing, they are elements constituting it. By contrast, having a gun in my bed stand, in my closet, or on my person might be a means for me to achieve my ends, but they are not constitutive elements of my flourishing. Hence, owning guns is not a fundamental interest.

Wheeler disagrees. He argues that the right to bear arms is fundamental since guns are the best way to protect our fundamental interest in self-defense (1997). However, on his view, guns are not inherently valuable; they are valuable only as a means of self-defense (pp. 433-8). I fail to see how this could make the right to bear arms fundamental. Not every means to a fundamental interest is a fundamental right. That would arguably make most actions protected by fundamental rights. Nonetheless, the connection between owning guns and self-defense is an important issue that I address later.

Others might claim that gun ownership is an essential element for the flourishing of a proper citizen. A proper citizen, on this view, is one capable of providing for and defending his family. Although each citizen can (generally) fend for himself, citizens come together to form a limited government to provide those few needs they cannot easily satisfy on their own. However, this vision of the citizen is very controversial, more controversial than the interest in gun ownership it seeks to justify. It assumes each of us has far more control over our lives than we arguably do have. Furthermore, even if this conception were defensible, it would not establish a fundamental right to bear arms since guns are mere means to independent citizenship; they are not constitutive of that citizenship. Hence, it is doubtful that the purported right to bear arms satisfies the first requirement of a fundamental right.

Second, we have evidence that granting this right type does harm society. If this evidence is at all credible, then granting this purported right would not satisfy the second requirement either.

But this does not resolve the issue. Although people do not have a fundamental right to own guns, gun control might be wrong because it violates some derivative right or simply because it is bad public policy.

(ii) Derivative right

Suppose we determined that "the right to bear arms" is not a fundamental right, but a derivative right. This would still be a significant finding since derivative rights, like fundamental ones, cannot be restricted without good evidence. Prima facie, I think we have such a derivative right. Each of us has a fundamental right of noninterference: we should be allowed to live our lives as we wish, so long as we do not thereby harm others. This is a right each of us needs, no matter what our particular interests. That general right derivatively protects personally important activities.

For instance, I would be furious if the state forbade me from sharing a pint with a friend. Nonetheless, although consuming alcohol is a particular interest and enjoyment I have, it is not a constitutive element of the good life in the way that the freedoms of speech, freedom, and association are. That is why I do not have a fundamental right to consume alcohol. Consequently, the conditions under which my consumption of alcohol can be legitimately restricted are more lax than they would be if the activity were a fundamental interest.

Nonetheless, since I have a prima facie derivative right to consume alcohol, the state can legitimately abolish or restrict alcohol consumption only if they can show that so doing is an effective means of protecting the public from harm. They can do that in some cases: people who consume substantial amounts of alcohol are dangerous drivers. Since this behavior is unacceptably risky to others, the state can legitimately restrict drinking while driving. Whether privately owning guns is similarly risky is something we must discover.

Bad Public Policy

If private gun ownership were not a derivative right, it might still be bad policy to substantially restrict or abolish guns. There are always costs of enforcing a law. Sometimes these costs are prohibitive, especially when the public does not support that law. If the public will not voluntarily comply with the law, then the state must try to force compliance. In their efforts to do so, they invariably employ excessively intrusive methods. Such methods never entirely succeed, and, to the extent that they do, they undermine public confidence in and support for all Law. Consider America's experience with Prohibition. Although one of Prohibition's aims - to protect innocents from harm caused by those under the influence - was laudable, the law was unenforceable and excessively costly. Consequently, less than two decades after Prohibition was passed via Constitutional amendment, it was repealed.

The cost of enforcing any law -- and especially an unpopular law - weighs against making any behavior illegal unless we have solid evidence that the behavior is seriously harmful. If we adopt a weaker standard - if we criminalize every action type whose tokens occasionally lead to some harm - then we would criminalize most behavior. Consequently, even if there were no right to bear arms, we should still not seek to substantially limit private ownership of guns unless we have good reason to think that will prevent serious harm.

Summing up: justifying the private ownership of guns

The preceding analysis isolates three questions we must answer in deciding whether people should be permitted to own guns: (1) How important is owning a gun to some people? (2) What are the consequences of private gun ownership? (3) Is abolishing or restricting private ownership of guns bad policy? Although gun ownership is not a fundamental interest, many people want to own guns and think they have good reason to do so. That is sufficient to show that serious gun control would undermine gun owners' interests. Moreover, there is some reason to think that serious gun control in countries with a strong tradition of gun ownership would be bad policy. Therefore, we should certainly not abolish, and arguably should not restrict, private ownership of guns without good reason. Are there good reasons? To answer this question, we must determine the effects of private gun ownership: (a) How likely is it that private gun ownership seriously harms others? (b) Are there substantial benefits of gun ownership that might counterbalance any harm?

Harm, Danger, and Risk

We must be careful when we say that guns cause harm. Guns kill people because agents use them to kill people (or misuse them in ways that cause people to be killed). As the National Rifle Association (NRA) puts it: "guns don't kill people, people do." In one sense their claim is uncontroversial: Murder is the act of an agent, and guns are not agents. In another way, their claim is irrelevant. No gun control advocate claims, hints, or suggests that guns are moral agents. Guns are objects and objects do no evil. But not all objects are created equal. Imagine the NNWA (National Nuclear Weapons Association) claiming that "tactical nuclear weapons don't kill people, people do." While in one sense their claim would be true, in a more profound way, it would be ludicrous.

Of course guns are not nuclear weapons. Guns are not as dangerous as nuclear weapons and some guns have seemingly legitimate uses. The question is whether the character of guns makes them especially harmful. We know that some objects - tactical nuclear weapons, biochemical weapons, live grenades, etc., are much more dangerous than feathers, ice cream, and butter knives. Where do guns fall along this continuum?

There are two distinct but related questions: (1) are guns "inherently dangerous"; and (2) what is the empirical probability that they cause serious harm? "Inherently dangerous" objects are those whose nature or design is sufficient to justify our prediction that they will cause harm, independently of any empirical evidence. We do not need double-blind empirical studies to know that nuclear weapons are inherently dangerous: they were designed to cause harm, and their nature is such that we can confidently predict they will cause harm. The two questions are intricately related since inherently dangerous objects are more likely to cause serious harm. Yet they are separable because some dangerous objects are not inherently so. Automobiles, alcohol, and cigarettes were not designed to cause harm, but all are causally implicated in many people's deaths. Other things being equal, we are more prone to control inherently dangerous objects than objects that merely have harm as an unwanted side-effect.

Guns, unlike autos, are inherently dangerous. Guns were invented for the military; they were designed to cause (and threaten) harm (Singer, C., Holmyard, E. J., Hall, A. R., & Williams, T. 1956: 367). The same aims determine the ways in which guns are redesigned: they are changed to make them more efficient at causing harm. In contrast, a significant aim of redesigning automobiles is to make them less dangerous. To some extent these efforts have succeeded. Although the absolute number of annual traffic fatalities has not noticeably declined, the number of fatalities per mile traveled has declined 75% since the 50s (Hemenway, D. 1995: 52.) We have enhanced the auto's original aim of efficient transportation while lessening harmful side effects. That is why we can sensibly say that the automobile is not inherently dangerous despite the fact that it causes harm. We cannot say the same for guns.

The literature of gun advocates supports my contention that guns are inherently dangerous. They advocate the private ownership of guns to prevent crime and to arm the militia. Guns can serve these purposes only because they are effective means of inflicting and threatening harm. Even guns normally not used to harm humans have purposes that ride piggy-back on this fundamental purpose. Shotguns are used to kill animals, and target guns are designed to be especially accurate. Taken together, this evidence supports the common view that guns are inherently dangerous. That is why we have special reasons to regulate them.

Although inherently dangerous, guns are far less dangerous than weapons of mass destruction, and they do have seemingly legitimate uses. That is why we must show just how risky they are before we can legitimately abolish or seriously restrict them. We must also determine if they have sufficient benefits so that we should permit them, even if risky.

An intermediate conclusion

We have shown that owning guns is not a fundamental interest and that guns are inherently dangerous. That is why we cannot categorically dismiss all forms of gun control. However, this is a weak conclusion. For although guns are inherently dangerous, they may not be so dangerous as to justify more than a system of minimal registration. What seems clear is that their inherent dangerousness precludes the idea that guns cannot be subject to governmental control. Some form of gun control cannot be categorically dismissed. Before determining the actual danger that guns present, we should first determine how risky an action must be before we can justifiably restrict it.

Risk

Humans are notoriously bad at judging risk. Often we are unaware of, or are inattentive to, the seriousness of risks. For instance, we may drive inebriated. At other times we overestimate the risks. For instance, we may refuse to fly because we think it is too dangerous. A proper determination of risk would be based on a careful accounting of the action's costs and benefits. We should determine (1) the probability of harm, (2) the seriousness of harm (the product of the gravity and extent of the harm), (3) the probability of achieving the benefits, (4) the significance of the benefits (the product of the importance and extent of the benefit), and then act accordingly. Of course even if we reached the same determination to the above questions, we might still disagree about whether to act: we might disagree about what risks are worth which benefits. Nonetheless, we can all agree that (a) as the likelihood and seriousness of harm increase, we have increased reason to refrain from acting, while (b) as the likelihood and importance of the benefits increase, we have increased reasons to act. We can import these lessons into the law.

Legal rules

But not straightforwardly. The issue is not whether we should own guns if they are legal, although that is a fascinating question. The question is whether the state should curtail private gun ownership. The foregoing considerations are relevant but not decisive. The decision to permit private ownership of guns is shaped by two factors pulling in opposite directions. First, even if we think Roger (an adult) stupidly engages in a dangerous activity (sky diving or boxing or racing), we might think Roger's autonomy requires that we permit it. Our commitment to individual liberty weighs against the government's abolishing or restricting the private ownership of guns as a way of limiting harm (Hughes, T. C. & Hunt, L. H. 2000). Second, some actions (smoking in public places) that are acceptably risky to Roger might be unacceptably risky to others. Are guns also unacceptably risky to others?

Put differently, gun control does not concern what private individuals should do, but what governments should allow private individuals to do. We must determine the risk of permitting the private ownership of guns, constrained by these complicating considerations. To illustrate how this might work, consider the following example. We have evidence that a number of wrecks are caused by drivers using cell phones. Roger wants to use his cell phone while commuting to work. He decides the inconvenience of not using the cell phone is worse than the small probability of personal harm. He might overestimate the inconvenience of not being able to use his cell phone or insufficiently appreciate the seriousness of the risk. However, since he is an adult, we might think we should not interfere with his decision to use a cell phone while driving. That is what autonomy requires. Yet Roger is not the only person at risk. Passengers in his or other cars may also be harmed. The seriousness of harm to them must also be considered in deciding to permit or restrict drivers' use of cell phones.

These judgements of risk must be further tempered by the costs of enforcement mentioned earlier. Although we know that using cell phones while driving may lead to accidents, we also know other activities may do the same. Drinking coffee while driving. Eating a donut. Looking at a map. Talking to a passenger. Driving more than two hours without stopping. Driving on less than six hours of sleep. Driving home after a bad day at the office. Presumably we should not make all these illegal. The probabilities of serious harm are small and enforcing such laws would require far-reaching intrusions into everyone's life. When the risks of an activity's causing grave harm to many others are small and the costs of interference are significant, then we should not criminalize the action. But as the probability of grave and widespread harm increases, then, other things being equal, we should criminalize the action.

For instance, when people are released from prison (and not just on parole) they have "paid their debt to society." Yet we do not permit them to own a gun. We judge that they are more likely to harm others. Of course not all of them - and likely not a majority of them - would harm others if they were permitted to own a gun. They are prevented from owning guns because they are members of a group statistically more likely to cause harm: we judge that allowing former felons to own guns is unacceptably risky. The National Rifle Association and most other gun advocates agree.

Someone might counter, though, that we deny felons the right to own guns not because we judge that permitting them to own guns is risky, but that they, by their actions, have forfeited the right to own guns. But that is not the best justification for our action. Why should felons forfeit their right after they have served their time and are free of all obligations to the state? For instance, while imprisoned in the United States felons do forfeit their right against unlawful searches and seizures. But once they are released from prison (and are no longer on parole or probation), a former felon has an unconditional right against unlawful searches and seizures -- the same as every other United States resident.

At first glance, there is some reason to think felons who use guns in commission of a crime could forfeit their right to own a gun, in the same way that drunk drivers lose their licenses. However, drunk drivers do not lose their licenses forever, while in most jurisdictions felons are never permitted to own guns. Moreover the prohibition against former felons' owning guns is not limited to those who used guns in the commission of a crime. Hence, it is more plausible to think that we can prevent released felons from owning guns because we judge that they are more likely to commit crimes with guns.

This is our rational for all laws proscribing risky actions. Every drunk driver does not cause an accident. Most do not. Yet we do not flinch at laws forbidding drunk driving. For it is not merely that drunk divers are statistically more likely to cause harm, they are more likely to cause harm because they are inebriated. We can arguably use the same rationale to justify restricting access to guns. We restrict access not only because guns are inherently dangerous, but because - if gun control advocates are right - permitting private ownership of guns is very risky.

 

What we Need to Know

We can now specify what we must know to intelligently decide whether to prohibit or restrict gun ownership (or any other risky action): (1) Is there a statistically significant correlation between the action (private ownership of guns) and harm (homicides, accidental deaths, suicides, armed robbery)? (2) Do we have good reason to think this correlation indicates that the purportedly risky action causes the harm? (3) How serious are these resultant harms? (4) How important is the activity that the state wishes to control (a) to the individual agent and (b) to the society?

In deciding whether to restrict the behavior, we must balance these considerations using the following general guidelines:

(1) If we have evidence that the behavior causes harm, then we have some reason to limit the behavior. As the evidence increases, the reasons for prohibiting the behavior increase. As the probability that the behavior will lead to serious harm (the product of the gravity and extent of the harm) approaches certainly, then the reasons for forbidding the behavior become very strong. (2) The more grave and widespread the potential harm, the more reason we have to constrain the behavior. If the gravity and extent of the harm are substantial, we might constrain the behavior even if our evidence that the behavior causes the harm is moderate. (3) The higher the probability that allowing the action will have important benefits, the stronger the reason to permit it. The greater the benefits, the greater the reason to permit it.

Libertarians might claim that individuals' rights are so strong that the state cannot justifiably intervene, even to constrain those who put others at extreme risk. The state should not proscribe risky actions, although they can intervene after harm has occurred. This use of "risk" is misleading. If on one occasion I drive while inebriated, I engage in a risky action: there is some probability that I and others will be harmed. However, permitting people to drive inebriated will definitely cause harm, although we cannot specify in advance who will be harmed. A personal decision to own a gun is risky in the former sense. A decision to permit citizens to privately own guns is - depending on the evidence - risky in the latter sense. If gun control advocates are right about the evidence, then we have good grounds to constrain private gun use. The question is: are they right?

Assessing the evidence

Armchair arguments

Debates over gun control typically begin, and sometimes end, with armchair arguments. Both sides offer armchair explanations of why (and how) the presence (or absence) of guns will increase (or decrease) violent crime. It is tempting to categorically dismiss armchair arguments since they seem to be poor substitutes for empirical evidence. However, it would be a mistake to assume we could devise sound empirical studies or understand their results without armchair arguments. In a study to discover if widespread availability of guns increases homicides or decreases crimes, we need armchair arguments to tell us which variables we should control (e.g., Lott, J. R. 1998: 21-4). Without them we would not know that we should control for the extent of poverty, the incidence of drug use, increases in the number of police officers, or the introduction of tougher (or more lax) penalties. Without them we would not know that we do not need to control for the price of mayonnaise, the criminal's eye color, or who won the World Series.

Armchair arguments also take center stage in evaluating empirical studies, in criticizing experimental design and in reinterpreting the reported findings (Black, D. & Nagin, D. 1998; Cook, P. J., Mollinoni, S., & Cole, T. B. 1995; Cook, P. J., Ludwig, J., & Hemenway, D. 1997; Hemenway, D. 1997; Hemenway, D. 1998; Lott, J. R. 1998; Wheeler, S. C., Jr. 1997). So before I discuss the empirical evidence, I summarize some significant armchair arguments employed by gun advocates and gun control advocates.

(i) More weapons, more violence

Gun control supporters offer empirical evidence of a positive correlation between murder rates and the availability of guns (especially handguns). Availability of guns is also positively correlated with suicide and accident rates. This empirical evidence is best understood against the background of the following armchair argument: (1) Guns (and especially handguns) are the easiest way to kill others or oneself. People can stand at a relatively safe distance and pull the trigger. (2) When people are angry, they act in ways they do not normally act. They may strike out at others. If they have a gun close to hand, they are more likely to use that gun. Although they could resort to a knife or a baseball bat, they are less likely to do so, and, even if they do, those weapons are less likely to cause a serious or fatal injury. (3) When people are depressed, they act in ways they would not act normally. If they have a gun close to hand, they are more likely to kill themselves. Although they might slit their wrists or take pills, they are less likely to do so, and, even if they do, they are less likely to kill themselves. (4) When people handle guns, even for a legitimate purpose, the probability of serious or fatal injury to themselves or others increases. When children have access to guns, the likelihood of an accident increases still more.

The conclusion of the armchair argument is clear: the more widely available guns are, the more people will be murdered, will commit suicide, and will die of accidents. This is a plausible armchair prediction. Perhaps it is wrong. Maybe it is reasonable but overinflated. Or it might be that the prediction is well-founded, but that the widespread availability of guns is nonetheless justified. What is apparent is that the claim that widespread availability of guns increases the number of homicides, suicides, and accidental deaths is highly plausible. It is difficult to imagine it is false.

(ii) Availability of guns prevents or stops crimes

Pro-gun supporters offer empirical evidence supporting the claim that guns prevent crime; their armchair arguments undergird and explain those studies. The motivating idea is simple: most criminals want to minimize their risks when committing a crime. If they know that someone in a house is armed, they will be less likely to enter that house, at least when the person is home and awake. Potential criminals are also less likely to assault or rob someone whom they believe is carrying a weapon. Finally, when criminals try to rob or assault an armed person, the person is more likely to foil the crime. This, too, is a plausible armchair prediction. Perhaps it is wrong. Maybe the claim is overinflated. Perhaps guns have these benefits, but there are other effects of owning guns - e.g., those mentioned above - which outweigh them. What is apparent is that the claim that the widespread availability of guns would prevent or thwart some crimes is highly plausible. It is difficult to imagine that it is false.

Of course we cannot stop with these armchair arguments. We must assess the empirical evidence.

The Data

The empirical evidence is difficult to assess, and, to the extent that we can, it does not univocally support either side. You might not know that from listening to the public policy debate. Some gun control advocates imply that strict guns laws would all but eliminate murder, while some gun advocates imply that having a gun in every home would virtually end crime. Both claims are unfounded. Gun control will not virtually eliminate murder. Arming all citizens will not virtually eliminate crime. About that we can be confident. The problem is determining the precise effects of permitting or restricting guns. The available evidence is less than compelling. But we must make a judgement based on the best evidence we have.

(i)The connection between availability of guns and murder

Perhaps the most well-established statistic is this: the more widely available guns (especially handguns) are, the more people are murdered. The figures are duplicated time and again in country after country. Here is the bottom line: "the correlation between any-gun prevalence and the overall murder rate is .67, while it is .84 between handgun prevalence and overall murder rate . . . " (Carter, G. L. 1997: 3). These figures are significant to the .01 level; that is, the chance that these correlations could occur merely by chance is less than one out of 100. This correlation meets the statisticians' gold standard.

But this does not resolve the issue, for it does not establish what gun control advocates claim it shows, namely, that gun control is an effective way of substantially lessening the murder rate. First, a statistical correlation shows that two things are linked, but it does not tell us if the first caused the second, the second caused the first, or if there is some third factor which caused both. Second, even if the items are causally related, we do not know that changing the cause will automatically and straightforwardly change the effect since another factor might intervene to sustain the effect.

Gun advocates proffer their own armchair explanation for the correlations: These correlations reflect the character of the respective social and political systems. The European countries where murder rates are lower have more social solidarity and are more heterogeneous than the United States. Whether these social factors explain all the correlation is debatable, but I am confident they explain some of it. Were the United States to regulate guns as tightly as most European countries, our murder rates would arguably fall, but they would not immediately plummet to their levels.

We might settle the issue if we conducted controlled experiments, randomly dividing our population in half, giving half of them guns, removing all the guns from the other half, and then monitoring the murder rate. Of course, that would be morally unacceptable, politically unrealistic, and probably even scientifically unachievable. Before we had enough time to exclude all possible intervening causes, sufficient time might have elapsed so that new intervening causes could have emerged. But we are not in the dark. We have empirical evidence that helps adjudicate between competing explanations of the correlation.

First, we have empirical evidence, bolstered by armchair arguments, that guns are more lethal than other weapons. Some claim the ratio is 5:1; no estimates are lower than 2:1 (Reiss, A. J., Jr. & Roth, J. A. 1993: 260). This partly explains the strong correlation between guns and homicides. If people get angry the same number of times, those using the most lethal weapons are more likely to kill their victims.

Second, the nature of secondary gun markets helps explain how the widespread availability of guns increases crime in general, and homicides in specific. Various opponents of gun control claim that "If we outlaw guns, only outlaws will have guns." Armchair arguments suggest why this is a silly claim. Where, one might ask, do criminals get their guns? They often steal them or buy them from those who purchased them legally. Even guns obtained from other criminals are usually traceable to people who purchased them legally. Empirical evidence supports this armchair supposition. Most criminals report having stolen their guns, received them from a friend or family member, or purchased them from someone who had stolen it. At least half a million guns are stolen each year (Cook, P. J. et al. 1995: 81), and these swell the numbers of guns available illegally.

Not only does the primary (legal) market effect the availability of guns on secondary markets, it also affects the price of guns on those markets, much "like the analogous markets for motor vehicles or prescription drugs" (Cook, P. J. et al. 1995: 71). As we restrict availability of guns in the primary market, the supply of guns in the secondary markets decreases and their cost increases (Cook, P. J. et al. 1995: 73). This increase in cost will diminish teenagers' ability to obtain guns, since they are least able to afford hefty prices. Since teenagers commit most deadly crimes, decreasing the availability of legal guns will thereby decrease the number of homicides. Conversely, having huge numbers of legally available guns increases the number of guns on secondary markets and typically lowers their price. This makes it easier for prospective criminals, including teenagers, to obtain guns.

Third, having a gun around the house (or on the person) - even for self-protection - apparently increases the chance that someone in the family will kill themselves with the gun, or will be the victim of a homicide or an accident. One study found that "for every time a gun in the home was involved in a self-protection homicide, they noted 1.3 unintentional deaths, 4.5 criminal homicides, and 37 firearm suicides" (Reiss, A. J., Jr. & Roth, J. A. 1993: 267). This implies that for every case where someone in a gun-owning household kills an intruder to thwart a life-threatening attack, nearly 43 people in similar households will die from a gunshot. Taken together the evidence does not prove that widespread availability of guns increases the number of homicides. However, that empirical evidence, bolstered by earlier armchair arguments, makes the claim highly plausible.

(ii) The use of guns to prevent crime

The biggest "gun" in the anti-gun control lobby is the claim that having (and perhaps carrying) a gun prevents crime. As I noted earlier, this is a sensible armchair claim. Someone contemplating a robbery is more likely to proceed if they think they can succeed with little risk to themselves. So if a prospective robber believes the tenants are at home and have a gun they know how to use, then he will likely seek another target. Two surveys support this belief. According to one survey, 4% of all Americans have used a handgun in the past five years to avert a crime. Given those figures, researchers estimates that there are at least 600,000 defensive uses of guns per year. Kleck uses these results, in conjunction with another survey, to claim that the number might be as high as 2.5 million (1991: 105-6). Given the number of violent crimes using guns, ". . . the best evidence indicates that guns are used about as often for defensive purposes as for criminal purposes" (Ibid. 107). If true, that is a powerful reason to resist attempts to limit availability of guns (Kleck, G. 1997). Such statistics, particularly when bolstered by moving anecdotes of those who have saved their lives by having a gun, cannot be cavalierly dismissed by gun control advocates.

However, these figures are inflated, likely dramatically so. First, Kleck's methodology is flawed. Surveys have an inherent tendency to overestimate rare events. Kleck made his estimates based on phone interviews with people in 5,000 dwelling units. One-percent of those units claimed to have used a gun defensively in the past year. Kleck inferred from these responses that there are 2.5 million defensive handgun uses per year. However, since this inference is based on an affirmative answer by one person out of a hundred, that means for every chance for a false negative (someone who falsely denies using a gun defensively) there are ninety-nine chances for a false positive (someone who falsely claims to have used a gun defensively) (Hemenway, D. 1997). The probability that this or some other bias skews the findings is substantial.

Second, Kleck's findings are inconsistent with findings by the National Crime Victimization Survey (1996), which interviewed far more people, and interviewed them more regularly. Kleck's estimates even clash with the NCVS findings on the incidence and circumstances of robberies (which seems less subject to reporting bias). If Kleck's figures were correct, then "Kleck asks us to believe that burglary victims in gun owning households use their guns in self-defense more than 100% of the time, even though most were initially asleep" (Hemenway, D. 1997: 1442).

Finally, if there were 2.5 million defensive gun uses each year, how many of those were necessary? Given the negative results of private gun ownership, gun advocates should show not only that guns deter crime, but that they are the best way of doing so. Some people plausibly claim that owning a dog is an effective deterrent. If true, then a not insignificant percentage of those who used a gun defensively could have achieved the same results without the accompanying danger. In summary, there is no doubt that guns deter some crime and stop the completion of other crimes. Just not in the numbers that Kleck claims.

John Lott supplements Kleck's argument by claiming that the widespread use of concealed weapons would decrease the annual number of homicides by 1,400, rapes by 4,200, aggravated assaults by 60,000, and robberies by 12,000 (Lott, J. R. 1998: 54). If true, and if there were no countervailing costs, this would be a powerful reason not only to permit guns, but to encourage people to have and carry them. However, Lott's conclusions have also come under severe criticism.

The central problem is that crime moves in waves, yet Lott's analysis does not include variables that can explain these cycles. For example, he used no variables on gangs, on drug consumption, or community policing. As a result, many of Lott's findings make no sense. He finds for instance, that both increasing the rate of unemployment and reducing income reduces the rate of violent crimes . . . (Hemenway, D. 1998: 2029)

Perhaps the most compelling critique comes from Jens Ludwig who compares the rate of violent crime toward youths and adults in states that passed shall-issue carrying permits. Most of these states issue guns permits only to people over 21. Armchair considerations predict that younger people, who cannot legally carry, will not receive the full benefits from the purported deterrent effect of shall-issue laws. Thus, those under 21 years of age are a natural control group to track general swings in crime. Once we include this factor, we find that shall-issue laws lead to higher - not lower - homicide and robbery rates (Ludwig, J. 1998).

I also have an overarching worry about Lott's conclusions. The one correlation in the gun control debate that is seemingly beyond dispute is the high correlation between the presence of guns - especially handguns - and homicide rates. Gun advocates offer explanations for the correlation, but no one I have seen seriously challenges it. I find it difficult to square this correlation with Kleck's and Lott's claims that having more guns - and toting them with us - will lower crime.

An overall assessment of the empirical evidence

The strong correlation between the presence of guns and higher murder rate is compelling. Since the correlation is statistically significant to a .01 level, it is difficult to believe that limiting private gun ownership will not have a noticeable effect on the number of murders. Gun advocates disagree: they claim that cultural factors explain the correlation. Although I think they are partly correct, they draw the wrong inference. For one crucial difference between European and American cultures is the widespread presence of guns. Each culture is the way it is, at least in part, because of the role guns (or their absence) played in its creation and maintenance. Therefore, curtailing the private possession of guns might well change the American culture so that it would be less violent. Consequently, it is not only that fewer guns would directly cause some decline in violent crimes - which it should. It is also likely to reshape the cultural values which, along with ready availability of deadly weapons, lead to such an extraordinarily high murder rate in America.

On the other hand, the statistical evidence that guns prevent or thwart crimes is suggestive and cannot be ignored, despite its identified weaknesses. In summary, the overall statistical evidence tilts in favor of gun control advocates, although the evidence is disputable. But we should not expect nor do we need indisputable evidence. We can act on the best evidence we have, while being open to new evidence. If widespread availability of guns were responsible for even one-fourth of the increase in the number of murders, that would be a significant harm the state should prevent if it could do so in a relatively unintrusive and morally acceptable way.

There is little doubt that we can do that, at least to some degree. If nothing else we could control some types of guns and ammunition. To take one obvious example, teflon-coated bullets are designed to pierce protective vests. People do not use these bullets to pierce the vests on a deer or a squirrel, on a target or a skeet. They use them to pierce the vests on people, usually law enforcement officers. This ammunition has no purpose except to cause harm. Hence, we are justified in abolishing teflon bullets and in establishing severe criminal penalties for those possessing them. This would not save large numbers of lives. But, assuming this ban's enforcement is not impractical, then, if it saved even a few lives, that would be a compelling reason to outlaw such bullets.

On the other hand, some guns have a much wider use, even if they are occasionally used for ill. People have seemingly legitimate uses for shotguns and single-shot rifles. Consequently, barring strong evidence to the contrary, we should not abolish them. We should, however, study their contributory role in causing harm, and explore ways we might lessen this harm in a relative unintrusive way.

The central debate concerns handguns. The evidence we have shows that handguns are disproportionately used in homicides and in robberies. Although "there are approximately three times as many long guns as handguns in the US, more than 80% of gun homicides and 90% of gun robberies involve handguns (Hemenway, D. 1995: 60). The experience in Canada suggests that criminals will not switch to long guns if handguns are unavailable. Given the special role handguns play in causing harm, we have compelling reasons to extensively control, or perhaps even abolish, handguns. But, policy considerations, mentioned earlier, should give us pause.

A third way

In the past we not only assumed that we must either support or oppose gun control, we assumed that the only way to control guns is to legally proscribe access to them. We should consider other options. Although I find the idea of a world without handguns immensely appealing, there are reasons to seek alternatives, especially in countries like the United States with a deeply entrenched gun culture. In the present political climate, the abolition or serious control of guns in the United States is unlikely to work and unlikely to happen. There are far too many people who desperately want guns. There are far too many people who own guns. Any attempt to disarm the society would be beset with the problems like those that plagued Prohibition. We have other possibilities.

We could employ elements of a policy we use to control another inherently dangerous object: dynamite. Dynamite has many beneficial uses. That is why we permit people to own it under specifiable conditions, e.g., to build a road. But it is also inherently dangerous. That is why we heavily restrict its purchase, storage, and use. I cannot own dynamite for recreation (I like the flash), for hunting (I am a lousy shot) or for protection (I would not hear an intruder). Owning dynamite is rarely a significant interest, and never a fundamental one. More important to the present point, even when we do permit people to own dynamite, we subject them to strict legal liability. The owner is financially liable for any harm caused by his dynamite, even if he was not negligent.

I propose we make handgun owners (and perhaps ultimately all gun owners) strictly liability for harm caused by the use of their guns. If Jones' child takes his gun and kills someone while committing a crime, then Jones will be financially responsible to those harmed. If Jones's child accidentally kills a neighbor's child, Jones will be financially responsible to the child's family. If someone steals Jones's gun and kills someone while robbing them, then Jones will owe the victim compensatory damages. And if Jones were negligent in the storing of the gun, he could be subject to punitive damages as well. Perhaps if he were grossly negligent in the storing the gun (he left if lying in his front yard, next to a school playground), we might even bring criminal charges against him.

This procedure is justified since guns are inherently dangerous, and it is only reasonable to expect people to take responsibility for their risky actions. The benefits are notable: many people would be disinclined to own guns, while those owning guns would likely take greater care in storing, handling, and using them. This could arguably achieve the central aims of gun control without direct government intervention. Doubtless that means that some people will be forced to pay for the misdeeds or mistakes of others in ways we might dislike. However, that is a more attractive policy than continuing the current scheme in which guns are easily obtained in the United States or than in completely denying individuals' interest in owning guns.

To make this option more palatable, we could let gun owners purchase liability insurance to cover potential losses. We might even require them to purchase insurance. After all, most states require drivers to have automobile insurance. This insurance-backed system of strict liability would make people take more care with any guns they own, while providing financial remuneration to those harmed by the use of those guns.

Perhaps this will not work. Other proposals might work better. What seems clear to me is that we need to do something: we cannot continue with the status quo.

Notes

^* I wish to thank Nicholas Dixon, Lester H. Hunt, Eva LaFollette, members of philosophy departments at East Tennessee State University and the University of Western Michigan, as well as the editors and two anonymous readers of this journal, for helpful comments and criticisms on earlier drafts of this essay.

Sources

Black, D. & Nagin, D. 1998: "Do Right-to-Carry Laws Deter Violent Crime?" The Journal of Legal Studies, XXVII (1), 209-220.

Carter, G.L. 1997: The Gun Control Movement. New York: Twayne Publishers.

Cook, P.J., Ludwig, J., & Hemenway, D. 1997: "The Gun Debate's New Mythical Number: How Many Defensive Uses Per Year?" Journal of Policy Analysis and Management, 16 (3), 463-469.

Cook, P.J., Mollinoni, S., & Cole, T.B. 1995: "Regulating Gun Markets." The Journal of Criminal Law and Criminology, 86 (1), 59-92.

Dworkin, R.M. 1977: Taking Rights Seriously. London: Duckworth.

Hemenway, D. 1998: "Review of More Guns, Less Crime." New England Journal of Medicine, 339, 2029-2030.

Hemenway, D. 1997: "Survey Research and Self-Defense Gun Use: An Explanation of Extreme Overestimates." The Journal of Criminal Law and Criminology, 87 (4), 1430-1445.

Hemenway, D. 1997: "The Myth of Millions of Annual Self-Defense Gun Uses: A Case Study of Survey Overestimates of Rare Events." Chance, 10 (3), 6-10.

Hemenway, D. 1995: "Guns, Public Health, and Public Safety." In D. A. Henigan, E. B. Nicholson, & D. Hemenway (eds), Guns and the Constitution. Northhampton, MA: Aletheia Press, pp. 49-82.

Hughes, T.C. & Hunt, L.H. 2000: "The Liberal Basis of the Right to Bear Arms," Public Affairs Quarterly.

Kleck, G. 1997: Targeting Guns: Firearms and Their Control. New York: Aldine De Gruyter.

Kleck, G. 1991: Point Blank: Guns and Violence in America. New York: Aldine De Gruyter.

Lott, J.R. 1998: More Guns, Less Crime: Understanding Crime and Gun-Control Laws. Chicago: University of Chicago Press.

Ludwig, J. 1998: "Concealed Gun-Carrying Laws and Violent Crime: Evidence from State Panel Data." International Review of Law and Economics, 18, 239-254.

Mill, J.S. 1978: On Liberty. Indianapolis: Hackett Pub. Co.

Reiss, A.J., Jr. & Roth, J.A. (eds) 1993: Understanding and Preventing Violence. Washington, DC: National Academy Press.

Singer, C., Holmyard, E.J., Hall, A.R., & Williams, T. 1956: A History of Technology. (Vol. II). Oxford: Oxford University Press.

United States Department of Justice. 1996: Criminal Victimization in the United States, 1993: A National Crime Victimization Survey. Washington, DC: US Government Printing Office.

Wheeler, S.C., Jr. 1997: "Self-Defense: Rights and Coerced Risk Acceptance." Public Affairs Quarterly, 11 (4), 431-443.