People who get fewer than six hours of sleep are more likely to be obese. People who talk on the phone in bed are more likely to get headaches. We hear about these "people who do this are more likely to be that" studies all the time; they're a staple of mass-media health reports, which are plucked from research journals. But they're among the most frequently misleading of all research studies, and for a simple reason: so many interconnected things are going on in people's lives that it's often nearly impossible to reliably determine that one factor is the main cause of some behavior, condition, or achievement. It may be true that a lack of sleep is linked in some way with obesity, but it's a big jump from there to conclude that if someone starts getting more sleep, they'll lose weight. It may be, for example, that people who sleep less also loosely tend to be people who exercise less, or eat less healthfully, or have a hormone disorder, or are depressed—in which case it could be any of these factor, rather than the sleep levels, that needs to be addressed in order to affect obesity. That would mean the link to sleep is pretty much incidental, mostly useless, and misleading.
— Wrong: Why experts keep failing us—and how to know when to trust them by David H. Freedman