Malcolm Gladwell has a blog

If you look, in fact, at emergency room statistics, you’ll see that more people are admitted every year for non-dog bites than dog-bites–which is to say that when you see a Pit Bull, you should worry as much about being bitten by the person holding the leash than the dog on the other end.

Does that follow?  Here is the blog, and thanks to Lynne Kiesling for the pointer.


Since you asked, I wondered the same think when I discovered MG's blog. And, no, I don't think it follows, because the statistics tell us that "more people are admitted every year for non-dog bites than dog-bites", not that between the people that see a Pit Bull more are admitted for non-dog bites than dog-bites.

That's pathetic and nonsensical. More bites are from not(x) than are from x. Without knowing how many not(x) there are compared to not(x) bites and how many x there are compared to x bites, the whole comparison is invalid (which is admittedly nothing new for him).

Following that logic, since you see more people (presumably) in an emergency admitted room for non-falling out of planes injuries than falling out of planes injuries, it's perfectly safe to fall out of a plane, assuming you don't get bitten by a dog on the way down.

More people die from car crashes than from car bomb explosions. So if you're driving along and you see a car full of explosives, don't step on the gas to try to get away--you should worry more about crashing than about the explosives going off...


Reminds me of the joke about blind skydivers.

Anyway, since we're piling it on this guy, how about this version:

"If you look, in fact, at emergency room statistics, you'll see that more people are admitted every year for non-dog bites than dog-bites--which is to say that when a snarling Pit Bull's jaws are inches away from your arm, you should worry as much about being bitten by the 90-year-old woman holding the leash than the dog on the other end."

No change in logical form.

Seriously, though, how many of those "non-dog bites" are hospital admissions solely out of disease concerns? 97% in the U.S. would not surprise me at all. Out in my neck of the woods, I've known several people bitten by raccoons, squirrels, or bats. Admission is precautionary, and many poor, uninsured people don't bother; obviously, the vast majority of those animals aren't carrying a dangerous disease. Dogs, especially a few notorious breeds, kill or maim a good number of people outright. Apples and oranges, really.

Also, on his website, Gladwell objected to the poor reviews Judge Richard Posner and I gave his bestseller "Blink." In particular, he was baffled and offended that Judge Posner and myself had scoffed at his claim that the reason car salesmen tend to offer higher prices to women and blacks is because the car salesmen are victims of their own unconscious prejudices, which keep them from making more money. Judge Posner and myself attributed this behavior, documented by Ian Ayres, to "statistical discrimination."

I particularly liked Gladwell's comeback: "Sailer and Poser [sic] have a very low opinion of car salesmen."

You can read all about it at:

Frequency of bites is not the issue at all. Pitbulls are jsut about average for NUMBER of dog bites, and labrador retrievers give the most bites, per year. The difference is that when a pitbull does bite you, it is more damaging than almost any other dog, and far more so than the human holding the leash.

Although it doesn't follow from his premise, Gladwell's statement is quite right -- you should fear the person holding the leash more than you fear a leashed dog. Dogs aren't very dangerous until they are unleashed.

"Blink" was an absolute travesty of a book. Gladwell was apparently inspired by the power of snap judgment, and wrote a 250+-page book whose grand conclusion was that sometimes it works well, sometimes it really doesn't, but apparently we should trust it more. Absolutely brutal; I think the low point was when he invented the concept of "temporary autism" to explain why people tend to make bad decisions when they are under a tremendous amount of stress and have no time to think, but I'm not sure. When I thought about how his book sold millions of copies because it appealed to people's snap judgments, I got angrier, although it would take me some time to explain exactly why (heh).

The book was amazingly vapid. It's fascinating that Steve Sailer briefly notes that ("go with your gut reactions, but only when they are right" -- dead on summary), and then goes into a horribly long screed about how Gladwell is a politically correct liberal. In particular, he is terribly upset that Gladwell assumes that the fact that car dealers charge blacks higher prices than whites is due to unconscious racial discrimination, when Sailer knows that the only possible explanation is that dealers are actually rationally playing the odds. In short, it was silly to assume that racism is irrational, because it is obviously the case that racism is super-rational, and everyone's feelings about race simply reflect a Bayesian analysis based on rap videos and the popularity of cognac in Detroit, among other things.

Actually, it's not that fascinating; more like completely unsurprising.

Sort of odd how people don't get it.. Gladwell's statement is nonsensical logic. (I'll even give gladwell some padding in my setup here)

Assume: More Human bites than dog bites.

Prove/Disprove: If you see a human and a dog together, the human is more likely to bite you.

The statement could easily be false. Of biting that occurs when a human/dog group encouters a solo human, who bites the most often? How often do dogs or humans bite when they are in a group? Maybe humans bite less when they're with a dog. Maybe dogs bite more when they are with a human (being protective).

If you make it a specific breed of dog, more questions come up.

His statement isn't obviously true or patently false.. It just exhibits his laziness.

One wonders whether his decision to write something so obviously wrong was a snap decision, or something he thought about for a while.

I was bitten by a dog as a small child (dog was not used to kids and I was harassing it). But we self treated it was pretty minor, while I've never been bit by a person that broke the skin I've heard that those are quite serious and should probably be treated by a doctor. I'm not sure that the number of emergency room visits for dog bites vs other bites is proportional to the actual occurance.
That said, I think there would be plenty of surprises in a study of reported bites by breed vs the number of dogs of that breed.

There's a similar mistake (or at least a similar lack of rigor) in the move from statistics to the risks of particular events in Freakonomics, when Levitt & Dubner discuss the risks that a girl would face if she went to the house of a friend with a swimming pool or to the gun-owning household of another friend. Levitt & Dubner at least make a minimal effort to correct for exposure, dividing deaths from the two causes by the number of guns and swimming pools in America, but this correction is obviously insufficient (click my name to see the discussion that begins with comment 10 to an unrelated Freakonomics post).

Comments for this post are closed