In 1995 a study in The Journal of Criminal Law and Criminology estimated that Americans used guns to defend themselves 2.1 million to 2.5 million times a year. That sounds like a lot, too much really—and the authors, Gary Kleck and Marc G. Gertz, acknowledge that, though they argue that it’s “not implausibly large” when you consider that there are 200-million-plus guns in the United States.
Nearly two decades later, that statistic has been recited countless times, and it often comes up in the aftermath of a horrible mass-shooting incident, like the one last week in Newtown, Conn., to make the case that on the whole firearms actually save lives and that gun-control advocates fail to see the big picture.
When Bob Costas recently argued, in response to a football player’s killing his girlfriend and himself, that “handguns do not enhance our safety,” the executive director of the National Rifle Association used that 2.5-million figure (it almost always gets rounded up) to slap him down with some peer-reviewed science.
But how good is that science?
The number has been questioned—ridiculed, really—by a number of other researchers, including David Hemenway, a professor of health policy at Harvard University and director of the Harvard Injury Control Research Center. In 1997, Hemenway, along with Philip J. Cook and Jens Ludwig, published a response to the 1995 study, calling 2.5 million a “mythical number.” In the Kleck study, based on a telephone survey of 5,000 people, the authors asked whether respondents had ever used a gun, even if it wasn’t fired, to protect themselves, someone else, or their property. If they answered yes, they were asked 30 more questions about the incident.
But there are multiple problems with the data, according to Hemenway and company. One is the likelihood that the respondent is not telling the truth, and that the gun wasn’t used in self-defense, but rather to commit a crime. If you’re involved in a drug deal gone bad, and you shoot someone, you might be more likely than a jury to chalk that up to self-defense—but giving researchers a false positive.
Another problem is something inherent in most survey research, a phenomenon called telescoping. If you ask someone about incidents in the last year, the stories they tell may have happened more than a year ago. The human memory doesn’t always line up neatly with the 12-month calendar.
And, Hemenway and his co-authors argue, in any survey a small percentage of answers are just going to be wacky. The person answering the phone may be drunk or senile or just lying. Do a phone survey asking about alien abductions, and you will get nearly a 1-percent positive response, which doesn’t confirm the existence of extraterrestrials. (Or does it?!?)
Kleck, a professor of criminology at Florida State University, responded to those criticisms at length in his 2000 book Armed: New Perspectives on Gun Control. Here’s an excerpt:
Hemenway presented a numerical exercise that, somewhat needlessly, confirmed the self-evident general point that estimates of rare phenomena could be highly sensitive to false positives, but also falsely claimed to have demonstrated that in fact the defensive gun use estimates from the NSDS [Kleck’s survey] were highly sensitive to, and greatly exaggerated by, false positives.
Why, Kleck wonders, doesn’t Hemenway address false negatives? If people lie about using guns to defend themselves, don’t others lie to conceal that they’ve used guns to defend themselves, perhaps fearing prosecution? By focusing only on the possibility of false positives, according to Kleck, Hemenway and his co-authors reveal themselves to be ideologues. He compares them to believers in so-called creation science who “speculate about the meaning of gaps in the fossil record” while ignoring the bulk of the evidence.
This debate started back around the time that today’s college freshmen were learning to crawl. Has there been any progress? Any meeting of the minds?
Nope. I called Kleck and Hemenway, and learned that the gulf today is as wide as ever. “Most of these things people claim are defensive gun uses are criminal gun uses in escalating arguments, which makes the world less safe,” Hemenway says. The 2.5-million figure is crazy, Hemenway says, and doesn’t line up with police reports and other statistics. He calls Kleck “not a very good researcher” and the “darling of the gun lobby.”
Kleck laughs at such descriptions. He says Hemenway has long had it in for him. When they met once, years ago, he says Hemenway refused to look him in the eye. “If he really was consistent in applying his theory, it would mean that survey estimates of the frequency of crimes committed by bad guys with guns grossly overestimate gun crime,” says Kleck. “Somehow he doesn’t get around to that conclusion, even though that’s where his logic leads.”
So where does that leave us? Here are a few things worth saying. One is that the 2.5-million number isn’t held out, even in Kleck’s paper, as a solid, indisputable number. It’s the high end of an extrapolated estimate, and it’s based on 20-year-old data. Citing it now, as the NRA and other gun advocates do, doesn’t make sense, even if you think Hemenway is wrong or biased. Also, it doesn’t—as even Kleck points out—indicate that 2.5 million crimes were prevented, even though that’s almost always the implication.
Also, citing that number doesn’t justify equipping firearms with large magazines or arming teachers. For the record, Kleck thinks arming teachers is a bad idea, and while he doesn’t think banning large magazines would make much of a difference in mass shootings (because perpetrators usually bring multiple guns), he doesn’t oppose it either. On those issues, at least, Kleck and Hemenway see eye to eye.