Editor's note: David Frum is a contributing editor at Newsweek and The Daily Beast and a CNN contributor. He is the author of seven books, including a new novel, "Patriots."
Washington (CNN) -- Do guns make us safer?
It's an article of faith among many gun owners that yes they do.
Last week, I presented in this space some evidence of the dangers of gun ownership: the elevated risks of accident and suicide in households that own guns. I pointed to a paradox: More Americans support gun rights, even as fewer Americans own guns. I explained this paradox with data that suggested many Americans hold false ideas about the prevalence of crime -- and wrongly look to gun ownership for self-defense.
Over the following seven days, I heard from many angry gun-rights supporters.
They argued that gun ownership is necessary for self-protection. They narrated stories of how their guns had saved them or their loved ones in armed confrontations.
And of course that must sometimes be true. The question is: How often is it true? And how do the benefits of widespread gun ownership compare with the measurable harms in higher rates of accident, suicide and crime?
Government figures from the National Survey of Criminal Victimization suggest 100,000 uses a year of guns in self-defense against crime, the vast majority of these uses being the display of weapons to deter or dissuade.
There are some problems with these government numbers, beginning with the fact that they are based on data from the early 1990s, when crime rates were much higher than they are today. The number of criminal attempts has declined 30% to 40% since then, and one would expect the number of occasions for self-defense to decline correspondingly.
For gun advocates, however, the main problem with the government estimate is that it is not nearly high enough to support their case that private gun ownership is the best way to stop crime. Many of them prefer another statistic, this from a study published in 1995 arguing that Americans use guns in self-defense some 2.5 million times a year, or once every 13 seconds. A Google search finds more than 1 million citations of this study posted online.
You can read the study here.
The trouble is that this claim of 2.5 million defensive gun uses is manifestly flawed and misleading.
Let's review the ways:
1) Even if you think the 2.5 million statistic was correct at the time it was computed, it must be obsolete today, for the same reason that the victimization survey data is obsolete. The 1995 study that generated the figure of 2.5 million defensive gun uses was based upon data collected when crime rates were vastly higher than they are today. Some of the data was collected in 1981, near the very peak of the post-Vietnam War crime wave. It's just incredible on its face that defensive gun use would remain fixed at one level even as criminal attempts tumbled by one-third to one-half.
2) When we hear the phrase "defensive gun use," we're inclined to imagine a gun owner producing a weapon to defend himself or herself against bodily threat. Not so fast. The authors of the 1995 study aggregated 13 prior polls of gun users, most of which did not define what was meant by "use." As the authors of the 1995 aggregation study themselves ruefully acknowledged: "The lack of such detail raises the possibility that the guns were not actually 'used' in any meaningful way. Instead, (respondents) might be remembering occasions on which they merely carried a gun for protection 'just in case' or investigated a suspicious noise in their backyard, only to find nothing." In other words, even if the figure of 2.5 million defensive gun uses had been correct at some point back in the early 1990s or early 1980s, the vast majority of those "uses" may be householders picking up a shotgun before checking out the noises in the garage made by raccoons rooting through the trash.
3) The figure of 2.5 million defensive gun uses is supposed to represent the number of such uses per year. Yet none of the studies aggregated in the 1995 paper measured annual use. Most asked some version of the question, "Have you ever?" Two asked instead, "Have you within the past five years?" The authors of the 1995 study took those latter two surveys, multiplied the rate in the survey by the number of U.S. households, then divided by five to produce an annual figure.
But people's memories of fixed periods of time are highly unreliable. It's not very likely that many respondents thought, "Today it's August 1990. I do remember scaring off a prowler in June 1984. But that was more than five years ago, so the answer to the question is 'No.' Not within the past five years."
More likely they thought, "I'll never forget the night I warned off a prowler with my shotgun. That was scary. Man, I'm glad I had my gun ready. When was that anyway? Three years ago? Four? I don't remember exactly, but the answer to the question is 'Yes.' "
4) Meanwhile, over in the world of hard numbers, the FBI counted an average of 213 justified firearm homicides per year over the period 2005-2010. If the figure of 2.5 million defensive gun uses were any way close to accurate, it would imply that brandishing a gun in self-defense led to a fatality only 0.00852% of the time. That seems almost miraculously low.
5) Underneath all these statistical problems is a larger conceptual problem. When we hear "defensive gun use," we're invited to think of a law-abiding citizen confronting a criminal aggressor. Yet crime does not always present itself so neatly. The vast majority of homicides take place between intimates, not strangers. Assaults, too, are often an acquaintance crime. When guns are produced by two parties to a confrontation, one party may deter the other. Yet it may be seriously misleading to designate one of these persons as a "criminal" and the other as a "law-abiding citizen." Perhaps when we hear "defensive gun use," we should not imagine a householder confronting a prowler. Perhaps we should think of two acquaintances, both with some criminal history, getting into a drunken fight, both producing guns, one ending up dead or wounded, the other ending up as a "DGU" statistic -- but both of them entangled in a scenario that would have produced only injuries if neither had carried a gun.
To be clear: I'm not disputing that guns sometimes save lives. They must. I'm certainly not disputing that the Constitution secures the right of individual gun ownership. It does. I'm questioning the claim that widespread gun ownership makes America a safer place. The research supporting that claim is pretty weak -- and is contradicted above all by the plain fact that most other advanced countries have many fewer guns and also many fewer crimes and criminals.
Should you own a gun? In some few cases, the answer to that question of wisdom is probably yes.
But most of the time, gun owners are frightening themselves irrationally. They have conjured in their own imaginations a much more terrifying environment than genuinely exists -- and they are living a fantasy about the security their guns will bestow. And to the extent that they are right -- to the extent that the American environment is indeed more dangerous than the Australian or Canadian or German or French environment -- the dangers gun owners face are traceable to the prevalence of the very guns from which they so tragically mistakenly expect to gain safety.
Follow @CNNOpinion on Twitter.
Join us at Facebook/CNNOpinion.
The opinions expressed in this commentary are solely those of David Frum.