The New York Times magazine has an article today entitled "Do We Really Know What Makes Us Healthy? (read it here)"
I'm impressed. Lots of great information on why we should be skeptical of epidemiological studies (you know, the ones that result in headlines like "Trans Fats Linked To Higher Risk of Irritable Bowel Syndrome"). I'll have to add to my list of statistics to beware of:
7. Be very skeptical of epidemiological studies that show a correlation between a certain behavior and a certain disease. Until there are an overwhelming number of them all showing a strong correlation, or until a double-blind, placebo controlled clinical study is done, they should be treated as an interesting hypothesis that needs more research.