Sam Walker was not your average American gun owner. For one thing, he had no interest whatsoever in hunting. And whereas the average gun owner owns at least three guns, Walker owned only one, a .38-caliber revolver, which friends persuaded him to buy for the sole purpose of protecting himself and his family in their suburban Houston home. Walker didn’t even particularly like guns. He still hadn’t gotten around to acquainting himself with his new weapon when his burglar alarm went off one weekday morning last December. Notified by his security company of the intrusion, Walker rushed home from work, quietly entered the house, took the gun out from the spot where he had left it for safekeeping, and, hearing a noise, moved stealthily up the stairs and opened a closet door. He saw a movement, a figure, and in a split second fired. The smoothly oiled gun worked perfectly, and Walker’s aim was true. A body fell to the floor. It was his 16-year-old daughter. She had cut school that day and had hidden in the closet to avoid her father. It wound up costing her her life.

If Walker’s tragic story argues against the benefits to be gained by gun ownership, consider an incident that happened a month later, across the country in New York City. One weekday morning in January, in front of a Brooklyn government building in broad daylight, Eric Immesberger stopped to give a man directions. Suddenly a second man came out from behind a pillar and knocked Immesberger to the ground. The two men then demanded his wallet and started beating him. Now, it just so happens that Immesberger is an investigator for the Brooklyn district attorney, and, more to the point, he was armed with a 9-millimeter semiautomatic handgun. He managed to pull his weapon and shoot one of the robbers in the chest. The other fled. Immesberger was later treated at a hospital for a broken nose.

Which case better represents the reality of owning a gun? It depends, of course, on whom you ask. But one point is indisputable: murder is committed more frequently in the United States than just about anywhere else in the developed world, and guns are its chief instrument. For African American males between the ages of 14 and 25, guns are the leading cause of death. And despite the recent downward blip in the numbers, crimes in the United States are far more likely to lead to death than they are in any other developed country. Every two and a half years, guns kill as many Americans as died in the Vietnam War. The litany of statistics is as deadening as it is depressing. Although few people would argue that cleansing the population of all guns wouldn’t go a long way to trimming the firearms fatality rate, the country’s 230 million guns, shielded by the Second Amendment, seem likely to remain in circulation for a long time.


Lacking a consensus on gun control, lawmakers have in recent years at least tried to put fewer guns in the hands of criminals and more in the hands of law-abiding citizens. The Brady Bill, for instance, seeks to curtail the proliferation of handguns, the weapons of choice for both crime and self-defense, by imposing background checks and a waiting period on new purchases. At the same time, the states are passing laws making it easy for residents to carry concealed handguns. But is arming the citizenry a good way to offset the risk of crime?

In the last decade researchers have focused unprecedented attention on the problem, and authors of some of the more dramatic studies have managed to amass impressively large stacks of press clippings. But science has not been especially helpful here. So far, nobody has been able to marshal convincing evidence for either side of the debate. “The first point that’s obvious in any scientific reading of the field is the extreme paucity of data,” says Franklin Zimring, a professor of law at the University of California at Berkeley. “What we have is critically flawed—on both sides.” Indeed, the scientific literature on the subject seems to teach very little, except for the tedious fact that it is difficult to say anything rigorously scientific about human behavior—particularly aggression.
What’s obvious by now to most scientists is that assessing the risk of owning a gun is nothing like assessing the risk of smoking cigarettes was 30 or 40 years ago. Back then medical researchers convinced themselves quickly of the cause-and-effect relationship between cigarettes and cancer. Although they had no direct, mechanistic proof, the epidemiological evidence proved the case far beyond any reasonable doubt. With guns, such a link has proved elusive, to say the least. Researchers think that about half of American households possess guns, they’re fairly sure that about two-thirds of these households have handguns, and they believe the proportion of handguns, within the total number of guns of all types, is rising. Their reasoning rests partly on the assumption that most guns bought these days are intended for self-defense; because of their small size, handguns are the overwhelming choice for this purpose. They also assume that the relative number of handguns owned will be reflected in the relative number of firearms deaths caused by handguns—about 60 percent.

Given the magnitude of the violence and the prevalence of the weapons, it is surprising that science has come to the issue of risk only recently. Criminologists have spent several decades exploring the impact of guns on crime and the behavior of criminals, but they have neglected the question of individual risk. When the medical profession got interested in guns in the early 1980s, it made them a public health issue, looking at the risk to the public at large. Emergency room doctors see the associated hazards every day, in the children who die or are wounded by playing with guns, in the successful and unsuccessful teenage suicides, and in countless other gun-related accidents claiming victims of all ages. The doctors concerned themselves not only with unintended firings but also with accidents such as Walker’s, in which the gun itself functions properly in only a narrow mechanical sense and the risk is more clearly seen in retrospect. And this public health perspective spurred renewed interest in studies that test to what degree the presence of guns increases the likelihood of death to their owners. But this approach, of course, focused on gun ownership as a societal issue; it did not assume the point of view of the individual. Doing so would have treated a gun as a consumer product, like a power drill or a lawn mower or a food processor, that carries with it a certain risk of accidental injury or death that must be weighed against its benefits.

Many of these public health studies attracted a great deal of publicity because they seemed to settle the question of risk once and for all. Arthur Kellermann, an emergency room doctor, is perhaps the most prolific and visible of the medical researchers who have tried to quantify the risk of owning a gun. Although he is a southerner who was raised with guns and who likes target shooting, he has nonetheless become a major source of bumper-sticker statistics for gun-control advocates. He insists that he has proved not only that a gun is a poor deterrent to residential crime but that having one actually increases the chance that somebody in your home will be shot and killed. In particular, his studies conclude that gun-owning households, when compared with gunless ones, are almost three times as likely to be the scene of a homicide and almost five times as likely to be the scene of a suicide. “If having a gun in the home was a good deterrent,” Kellermann says, “then we should have seen few guns in the homes of murder victims. But we found the opposite.”

Kellermann’s work has drawn fire from researchers who suspect that his passion for the issue has blinded him to ambiguities in his data. “Kellermann has decided that guns are bad, and he’s out to prove it,” says Yale sociologist Albert Reiss. Although in general criminologists don’t object to Kellermann’s research methods, they part company in their interpretation of his results. His evidence, say critics, is so riddled with uncertainties as to preclude any definitive interpretation.