Gary Taubes had a great rant in March on misinterpreting correlation in health research. He describes the compliance effect, where the “Girl Scouts” in longitudinal health studies do everything they’re told is good for their health. So they are very healthy, but this doesn’t mean that every single thing they do is good for your health. It’s a great point to keep in the back of our minds when we attempt to attribute an effect, particularly in longitudinal studies.