Upon close inspection, kellermann’s results are much more modest than his dramatic conclusions would indicate. He chose to study guns in the home not only because lots of people buy them for self-defense and keep them in a drawer beside their beds but also because home is a well-defined place that simplifies the task of collecting data. Police homicide records specifically include the location of each incident and the weapon used, and it was a straightforward matter for Kellermann to follow up each case by interviewing surviving family members and friends. The problem was in coming up with a suitable control group against which to draw comparisons. Ideally, you want to pair each victim with a control that differs from the victim only in that one was shot and the other wasn’t. Kellermann devised a clever methodology for doing so. For each victim, he randomly selected one neighbor after another until he found someone who was the same age, sex, and race. Eventually he assembled “matched pairs” for 388 homicide victims.
When he compared the victims with the control group, however, he found that many more factors differentiated the two groups than their victim status. It turned out that the households in which homicides took place were more likely to contain a family member who abused alcohol or drugs and had a history of domestic violence—these factors contributed to the likelihood of homicide independent of the existence of guns.
Kellermann took pains to compensate for these other factors using standard statistical techniques of epidemiology. In essence, he tried to estimate how much each factor, such as alcohol abuse, might have influenced the homicide rate among victims in his study, and then he adjusted his figures accordingly.
What neither Kellermann nor his critics can know for certain is whether this statistical juggling actually uncovers any underlying trends or whether something else is going on that Kellermann hasn’t accounted for. Kellermann himself admits the possibility of some kind of “psychological confounding”—that some intangible factor such as aggression, rather than merely the presence of guns, is influencing the results. Critics also point out that the victims in Kellermann’s study may have gotten guns because they felt themselves to be threatened in some way, which means they might have suffered higher homicide rates even if they hadn’t bothered to arm themselves. “Kellermann has shown that homicide victims are more likely to keep a gun at home, but criminologists have known that for years,” says Gary Kleck, of Florida State University in Tallahassee.
The results of one survey, to find out how often guns are used in self-defense, depict the country’s gun owners as holding back a tidal wave of violence and crime.
Kellermann’s even more dramatic figures on suicide in the home are especially problematic, mainly because Kellermann relies on the numbers without offering an explanation. “There’s no theory to account for his conclusion,” says Zimring. Suicide is also thought to be prone to substitution—that is, although guns are the preferred instrument of suicide in the United States, a person bent on suicide can easily find a substitute if need be. Since Kellermann’s study focuses on suicides in the home, it doesn’t account for the victim who, lacking a gun, decides instead to jump off a bridge.
Regardless of their personal feelings on guns, criminologists, who tend to look at violence through the lens of police statistics and surveys, are usually more open than doctors to the possibility that a gun can now and then deter a crime. Trouble is, social scientists are poorly equipped to measure events that do not occur—crimes that are averted because the would-be victim had a gun. As a result, criminologists have resorted to surveys to get at this phenomenon. Most recently, Kleck conducted a survey to find out how often gun owners actually use their guns in self-defense. His controversial results depict the country’s gun owners as holding back a tidal wave of violence and crime. He estimates that 2.5 million times each year, somebody somewhere in America uses a gun in self- defense. This figure has become a mantra of the National Rifle Association (with whom Kleck has no affiliation).
Most other criminologists are critical of Kleck’s methods, and almost all of them are incredulous at the results. A big complaint is that he leaves it to his survey respondents to define a “defensive gun use,” so he may have captured incidents that most people would consider trivial. “An awful lot of what some people would call self-defense is, like, somebody asks you for a quarter and you tell them to get lost, but as you walk away you keep your hand on your gun,” says Philip Cook, a Duke University economist. In addition, many incidents that people report as self-defense may in fact be assaults, in which the respondent takes a more active role than he admits. “In many instances, we may only be talking to one side of an argument,” says Zimring.
What this criticism comes down to is that Kleck, like Kellermann and all the other researchers in this field, is guilty of failing to explain what happens when people carry guns, and how possessing one affects their interactions with criminals. As Reiss puts it, “We know very little about how motivation enters into an action.” Zimring likens efforts to understand the deterrent effect of guns to “dancing with clouds.” Kleck himself admits that “the better the research, the more it tends to support the null hypothesis—that gun ownership and control laws have no net effect on violence.”
Even when a seemingly perfect opportunity for a real-life experiment presents itself, as it did recently to criminologist David McDowall, the null hypothesis is often all that a criminologist is left with. Several years ago, Florida, Mississippi, and Oregon adopted “shall issue” laws requiring the states to issue a license to almost anybody who wants to carry a concealed handgun. McDowall saw that the effect of these laws would give him a laboratory in which to test the arms-race hypothesis: he could find out whether criminals, knowing their victims are more likely to be armed with handguns, are more likely to use guns themselves. He could also find out whether citizens, when armed, can deter crime.
After the new laws were passed, permits to carry concealed handguns rose enormously —in Florida the number of licenses soared from 17,000 before the law was passed in 1987 to 141,000 seven years later. After studying five cities, McDowall found that the rate of firearms homicides increased overall by 26 percent. Although this would seem to support the arms-race hypothesis, the results were inconsistent. Whereas McDowall had expected the effects of the liberalized laws to be greatest in Miami, the biggest city in the study and the one with the highest crime rate, the rise in homicides there was too small to be statistically significant. However, McDowall believes his evidence is strong enough to show that armed citizens do not decrease the number of firearms-related deaths.
Despite the frustrating lack of clarity, researchers are universally optimistic that, with time and the accretion of data, insight into the mechanism of violence will come, and with it, a greater consensus on the real risks of guns. For the time being, however, there will remain very little one researcher can say about risk that another researcher cannot refute. Most favor restricting the availability of guns by mandating background checks and waiting periods, which serve to some degree to keep guns out of the hands of “hotheads” and criminals. There is also a consensus that higher homicide rates have everything to do with the preponderance of guns—an obvious inference when considering, say, crime statistics of London and New York. These two cities have similar crime rates, but the homicide rate from burglaries and robberies in gun-rich New York is vastly higher—54 times higher in 1992, according to Zimring. “America doesn’t have a crime problem,” he says, “it has a lethal violence problem. It’s that thin layer of lethal crime that Americans are afraid of.”
Given that purging guns from the population is problematic, would the world be safer if each law-abiding citizen carried a gun? Alessandro Veralli hesitates before answering this question. For most of his adult life, he has carried a concealed handgun almost everywhere he goes, whether it’s out to the movies with his wife or to the local hardware store on a Saturday afternoon. Yet Veralli, a Master Firearms Instructor for the New York City Police Department and an NRA life member, admits that as a civilian he has had very little opportunity to use his gun. If he ever found himself a customer at a liquor store that was being held up, in most cases his training and common sense would tell him to lie low rather than start a shoot-out. If he was out with his wife and a thief demanded his wallet, he would probably hand it over. “In a robbery, there’s not much you can do except maybe shoot at the guy as he’s walking away,” he says. “But what if he shoots back? I’d be putting my wife in danger, and for what?” He carries a gun for the hypothetical extreme case when having it might mean the difference between life and death. “Personally I’d hate to get into a bad situation and think that I might have been able to do something if I had had a gun,” he says.
But should other citizens carry guns? “I’m tempted to say yes,” he says, but then he demurs. “Maybe it makes sense in other parts of the country where they have more space. New York, though, is too crowded. There’s something about all these people being confined in a small space. People can fly off the handle over little things. I don’t think I’d want to see each and every one of them carrying a gun.”