Confusing article on gun statistics

Today’s New York Times carries an article about a new survey of gun ownership in the United States. The main drift is that there has been a strong shift away from the ownership of long guns in favor of handguns, and that people are buying handguns for self defense. The comparison is between 1994 and 2015.

The graphs, as we’ve come to expect from the Times, are very attractive. But some of them are unclear.

I can’t tell from the graph reproduced here, for example, what the numbers are supposed to mean. Look at the first number, for male handgun ownership (22%). Does this mean that 22% of all men own handguns? Or does it mean that 22% of all handguns in private hands are owned by men? Or, perhaps, that of men who own guns, 22% of those guns are handguns? Or something else?

I have no idea. Journalists are famously bad at dealing with anything quantitative, but usually the interested reader can check the original paper to learn the methodology and get the real story. Not here, though. This research won’t be published until next year, but its authors shared the results in advance with the newspaper, which put them to the usual purpose: as an elaborate preface to trotting out the 1993 NEJM study that purports to show that a gun in the home increases the likelihood that its residents will become victims of suicide or homicide. I am pessimistic that the journalists in this case are savvy enough to avoid the statistical fallacies that would lead one to the conclusion to which they are so eager to lead the reader: that population averages somehow mean that any individual who keeps a gun in the house is more likely to die from gun violence thereby.