Doctors and Statistics

by F.

Seth comments on Dr. Schwartz’s contention that statistics are “totally meaningless when applied to the individual.”

Oddly, a doctor friend of mine has said the exact same thing to me many times, making me wonder if this bit of dogma isn’t part of the medical school curriculum. In the textbook Social Psychology written by Timothy Wilson et al, there’s a nice graphic showing that doctors are right up there with law students in statistical sophistication (And can you guess how bad most law students are with statistics?)

Once, when my doc friend and I were eating burritos together, he restated his “statistics don’t apply to individuals.” I gave him the “lottery ticket” response: “So, if I buy a lottery ticket and the odds are one in two million that I’ll win, those odds don’t apply to me?” His response: “I’ll have to think about that.” The next time he said this I gave him the “plane crash” response. “So, if the odds of my plane crashing are 0.00001, then I shouldn’t get on the plane because—screw the odds. It could be me! This could be that one plane!” Again, disagreement but no response.

There is a truth in the vicinity, of course: it’s that evidence about an individual trumps evidence about a statistical population of which the individual is a member. For instance, if the average age of people in your country is 27, that doesn’t mean that you’re 27. Or there’s the example about testicles and ovaries: on average, Americans have one testicle and one ovary. Or the barnyard example: put a chicken and a cow in the pasture. Residents of the pasture now have, on average, three legs. This hardly means that “statistics are meaningless when applied to the individual,” nor does it mean that “anyone can lie with statistics.” Stats are a tool. I mean, if I give you a ball peen hammer, you could: (a) shot-peen some metal, or (b) hit yourself in the face with it. One use is slightly more sensible than the other.