I edit for a scientific journal, and I've been doing it for over 11 years (see previous entry). I can pick apart epidemiological studies with the best of them, so I pulled up the New England Journal of Medicine article about how social networks facilitate the spread of obesity to see how the authors reached this conclusion.
(Note: that link is to the full-text version, so anyone can read the whole damn article themselves.)
The article tells us nothing. I mean that. NOTHING.
The only data in the article tell us this: friends of fat people have a greater tendency to become fat themselves than do the friends of thin people. It doesn't tell us WHY. Essentially, the article says, "Look, here's a connection!" That's normally the starting point for a study. Scientist sees a connection, and then tries to figure out what causes the connection. "Figure out," in a scientific setting, does NOT mean "guess." It does not mean "assume." It means "generate reproducible facts and data."
This article did NOT generate reproducible facts and data to explain this connection. Basically, the upshot of the entire article is that the authors think that, merely by being fat, fat people make their friends believe it's okay to be fat. I mean, seriously. Unless there's a significant portion of the article not present on NEJM's Web site, that's the only conclusion the article makes.
That's NOT evidence; it's merely reporting of a correlation. CORRELATION IS NOT CAUSATION. And I know that NEJM knows damn well that that's so.
And it pisses me off mightily that the mainstream media grabbed this utterly specious piece of crap and ran with it, just to fan the flames of fat hatred. That's just WEAK.