If you can't trust the results of a self-selected group of individuals who are paying a good chunk of cash for the express purpose of loosing weight... who the hell can you trust?
Generally speaking you're absolutely right, but with diet the social pressures are significant enough to compromise analysis even if the group is paying for it. There have been quite a few studies on the topic, e.g.: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.103...