[tsimionescu]

> One the other hand, this is how science works. We do experiments, get the results of those experiments, and try to make the best use of the data we have.

That is not how science works. A scientific experiment has to have enough statistical power for the claim you are trying to make. If it doesn't, then you don't make the claim.

As you say, it's extremely hard to make good (I would say even decent) nutritional studies. The vast majority of nutritional studies are not even close. Why then do dieticians pretend to be medical doctors and give advice based on these studies? Why do they give advice to entire nations based on scientific-sounding gobbledygook?

When you don't know something, the scientific thing to do is to say you don't know. Physicists don't claim they know how gravity works at the quantum level just because they have some models that sorta kinda work sometimes. A linear model of a non-linear phenomenon is not in any way better than no model - it's obviously worse.