Newsprint reporters are probably a dying breed and should be treated gently. So, I will simply note that Nanci Hellmich of the USA Today in an article today seems not to understand the basic principle of scientific experimentation.
She leads this article with the following. "Dieters who write down everything they eat each day lose twice as much weight as those who don't, according to one of the largest weight-loss studies ever conducted. This confirms the importance of keeping a food diary." The only real confirmation here is this passage confirms the error in the title "Using food diaries doubles weight loss, study shows.
This is simply wrong. The dieters were not randomly assigned, this was just the first phase of an experiment which in phase two will test, randomly assigned, subjects who already lost weight to a number of maintenance scheme. The design is to see what maintenance schemes work.
The doubling of weight loss, referred to in the article, is simply that the loss for those who by self selection used a food diary is twice that of those that did not. Obvious confounds include the possibility that those who kept diaries were more motivated to lose the weight, or those who were losing weight were proud of their achievements and motivated to keep a diary.
The article is by J. Hollis and others, published Aug 2008 in the American Journal of Preventive Medicine, Vol 35 (2) pg 118-126.
Clearly stated in the article itself is: "A limitation of Phase I of WLM was that it was an uncontrolled observational study." Exactly.
Cause and correlation are difficult to separate. That is the entire reason for random assignment controlled experimentation. Well controlled experiments are still sometimes wrong, correlations picked out by observation are much much worse. Confusion between the two is disastrous in the attempt to explain scientific work to the public.