I think the trouble is what they leave uncontrolled. If type of food is important and you only measure total calories you have a problem in your experiment desi

I think the trouble is what they leave uncontrolled. If type of food is important and you only measure total calories you have a problem in your experiment design. It seems here there is also a subject selection problem. A 5% body fat difference for starting out is pretty significant. This is the trouble with the vast majority of this research. It's done in such a way that it's hard to learn much from it.