A recent Czech study about a supposed inverse correlation between beer-drinking and scientific output (the more beer you drink, the less productive you are - read the NY Times report) has been the butt of jokes around the scientific blogosphere: scientists drink a lot of beer, even successful ones. Most scientists probably would have guessed that there is a positive correlation between success and beer-drinking in science. So what's the deal? I thought the data didn't look so impressive - a fairly small number of data points, not a great correlation coefficient (but not terrible either). But I didn't look too carefully - it's going to take a lot more than 34 Czech ornithologists to convince me to give up beer. It turns out the big problem in the study was the presence of a few outliers: apparently, "the five worst Czech ornithologists drank a lot of beer". If you get rid of the five outliers and look at the remaining 29 ornithologists, there is no correlation. I'm glad we don't have to cancel our department happy hour.