undefined | Better HN

0 pointsjacques_chester11y ago0 comments

That meta-analysis was not based on randomised controlled trials. It was largely based on observational and self-reporting studies, which are notoriously inaccurate.

Moreover, it was subject to numerous corrections due to sloppy work by the authors:

http://news.sciencemag.org/health/2014/03/scientists-fix-err...

0 comments

kareemm11y ago

Thanks, hadn't seen this link. I'd like to see the specifics of what the errors are. Agree that observational and self-reporting are inaccurate, though unfortunately they're some of the best data we've got wrt nutrition.

Also, looks like there were some random trials in the meta analysis. From your link:

"These included trials in which participants were randomly assigned to different diets, as well as observational studies in which participants' intake of fatty acids was determined by asking them about their diet or by measuring the fatty acids circulating in the bloodstream."

jacques_chesterOP11y ago

> 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.

kareemm11y ago

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.

1 more reply

j / k navigate · click thread line to collapse

0 pointsjacques_chester11y ago0 comments

That meta-analysis was not based on randomised controlled trials. It was largely based on observational and self-reporting studies, which are notoriously inaccurate.

Moreover, it was subject to numerous corrections due to sloppy work by the authors:

http://news.sciencemag.org/health/2014/03/scientists-fix-err...

0 comments

kareemm11y ago

Also, looks like there were some random trials in the meta analysis. From your link:

jacques_chesterOP11y ago

> 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.

kareemm11y ago

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.

1 more reply

j / k navigate · click thread line to collapse