Another Horse Laugh

Peter Austin's argument on data mining (earlier discussed here) has been made even more concisely by James Lileks (on a study that purports to show that chocolate can be good for the arteries):

Someone has a good job spending grant money to try all sorts of arbitrary combinations. You never hear about the studies that fail. Eggs do not prevent those weird super-strength hairs in the ear; gargling Spam smoothies has no measurable effect on liver function, although it seems to strengthen the gag reflex. And so on. Right now, some researcher no doubt is assembling a proposal to test the effect of Pixy Stix on nasal congestion. (They seem to work, if you push hard enough.)