Or for one of my favorites, 'a glass of wine a day is healthy' because of antioxidants... or something. Except it isn't, at all. But rather as people's health deteriorates to a sufficient degree they tend to stop drinking (even if not consciously because of their own health), and so when you study light drinkers versus non-drinkers, there was enough of an undetected bias in there to make it seem that drinkers were, on average, healthier. It's easy to see echoes of that possibility in near to any study, especially observational.