Now the gun controllers pour out of the woodwork to claim that you're more likely to kill yourself or a family member with a gun than a criminal.

Some of the people deploying this statistic really ought to know better. Composition fallacy, anyone?

These are not double blind experiments. Guns may be the weapon of choice for all sorts of crimes; that does not mean that they cause the crimes.

Men like to kill themselves with guns. (This is not culture-specific; women tend to choose poison everywhere, presumably because of some deep fear of disfigurement). Gun suicides tend to be successful. But this does not mean that if you took away the guns, people wouldn't commit suicide. There are many other near-surefire ways of killing yourself, like jumping off a high bridge, gassing yourself with carbon monoxide, driving your car at high speed into a piling, hanging yourself, etc. Think of it this way: most people who choose to wear high heels are women. That doesn't mean that if I threw out my Manolos, I would turn into a man.

Similarly, (a small number of) men like to murder their families with guns. But they also like to murder their families with knives, baseball bats, and their fists. Taking away the guns might somewhat reduce the number of homicides (it might also increase it; you're more likely to recover from a fatal-looking gunshot wound to the stomach than from having your head banged against the floor 80 times). But spousal murder is plenty easy without a gun.

Now compare this to the actions of people who are not looking to commit homicide or kill themselves. What are they likely to do with a gun? Brandish it or fire a warning shot. If they do shoot someone, they are likely to stop as soon as that someone is disabled, and call for an ambulance.

Most importantly, almost no homicides or shootings go undetected. On the other hand, many people who wave a gun at someone threatening, and thereby cause that person to go away, don't report it. How many? We don't know, because they don't report it. But I didn't report the mugging I foiled last January through strategically hunting for my keys in a well lighted portion of the street. I doubt I'd have been any more likely to do so if I'd waved a gun at him.

Now, it is possible that having a gun is actually on net dangerous to you and your family. But we have no evidence to support this notion, because all the statistics on the subject are crap. The denominator is what criminologists call a "dark number": one where there is no way to arrive at any reasonably credible estimate of its value.

Indeed, there is a thesis to be written, somewhere, on why so many of the most famous works on guns involve statistical malpractice, deliberate distortion, or outright lying.

Look at the good longitudinal data we do have: liberalizing concealed carry--the right to have a hidden gun on you at all times--hasn't resulted in the predicted rash of deaths. It turns out that suicides and would-be homicides weren't paying much attention to the legality of their actions. It also turns out that having a gun in your hands does not seem to turn a previously law-abiding citizen into a spur-of-the-moment killer. It wasn't totally unreasonable to fear that guns might turn altercations into homicide, but the evidence from states that have moved to shall issue models shows that they didn't.

Don't listen to me, gun nut that I am; listen to liberal crime policy professor Mark Kleiman. People who shoot other people, or themselves, are not ordinary folks whose gun let them vent a moment of madness. They're mostly people with long histories of all sorts of violence towards either themselves, or others.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.