[jacques_chester]

> Also, looks like there were some random trials in the meta analysis.

There were, but the statistical effect is drowned by the much larger pool of observational/self-reporting studies.

Observational studies are a useful place to find potential leads, but they are nowhere near as reliable as RCTs for many reasons. Observational studies can tell you that the fall in piracy is causing global warming. Short of a randomised controlled trial involving the introduction of pirates or non-pirates at random to 13,000 Earth-like planets, it's hard to know how strongly that relationship holds.

[kareemm]

I agree with you about observational studies when I wrote in the grandparent:

> Agree that observational and self-reporting are inaccurate, though unfortunately they're some of the best data we've got wrt nutrition.

[MDS100]

They are not the best data. It's just too widely used.