Can this study possibly be as stupid as it sounds?

This University of Pennsylvania study says that having a gun doesn't help when you're assaulted.  In fact, it makes things worse.  Much worse:

In a first-of its-kind study, epidemiologists at the University of Pennsylvania School of Medicine found that, on average, guns did not protect those who possessed them from being shot in an assault. The study estimated that people with a gun were 4.5 times more likely to be shot in an assault than those not possessing a gun.

That's newsworthy.  It's both counterintuitive and supports conventional liberal views of guns.   What's not to like?

Well, look how they came up with the statistic.  First, they found people who had been shot in Philadelphia and discovered that six percent were carrying a weapon when they were shot.  Then they compared those people to "random Philadelphians" called by "trained phone canvassers."  These "random Philadelphians" turned out to be 4.5 times less likely to own a gun than the people who got shot. 

So let's suppose you lived in Philadelphia, in a really bad neighborhood where you thought you were likely to get shot.  Would you be more likely than a "random Philadelphian" to buy gun -- and to have it on you when, as feared, you got shot?  Yup.  How about women who've been threatened by angry ex-lovers?  More likely to get a gun?  More likely to get shot?  Yup and yup.  Same for drug dealers.  Same for police officers.  All more likely to get shot than random Philadelphians, and all more likely to have guns.  So, does this study actually tell us anything?  Hasn't Charles Branas, the author, simply made a remarkably stupid error in choosing the control population? 

I can't find the study on line, only the university press release, so maybe there's a good answer, though releasing the press release without the study isn't comforting.  And even the press office seems to have realized there might be a problem.  So they put on their white coats and tell us not to worry. We're the stupid ones.  You see, they tell us, "This is the same approach that epidemiologists have historically used to establish links between such things as smoking and lung cancer or drinking and car crashes." 

Sure, that makes sense.  Same thing exactly.  I've been thinking about buying some cigarettes to protect against attackers wielding lung cancer, but first I've got to lay in a supply of liquor to fend off the drunk drivers.

Laugh if you like, but the study is being published by the American Journal of Public Health -- the same people who are supposed to protect you from H1N1.