A. This is one guy, basically. I mean, I agree with him, generally, but not everyone does.
B. It's easier to poke holes in someone else's thing than create a good study.
So study A: "I wonder how fat you have to be to not fit through this door. Let's weigh people (instead of measuring them) to see how fat you have to be not fit through the door."
Study B: "It's a great question, but it's badly done. Weighing them won't work. Let's just interview people and write down their self reported height!"
Meta analyst: "Y'all are both wrong and stupid, geez."
Scientists A and B: "Let's take the meta-analyst out back and beat him up!" "Sounds good to me!"
This is not a stupid question. Many meta studies are done by people with their own frameworks of deciding what is and isn't a legitimate approach to studying something. Unsurprisingly, their findings are overwhelmingly negative due to their own publication bias.
Meta studies are pretty worthless in my opinion. Primary studies in all fields should be forced to be published with open data, and any study that received - directly or indirectly - more than $1 from the public taxpayer should be forced to publish Open Access.
Lay bare all the raw data and let the software developers and data analysts get after it, not so-called expert meta reviewers.
Really can't tell if you're being sarcastic here, even in the context of HN.
> Meta studies are pretty worthless in my opinion. Primary studies in all fields should be forced to be published with open data ...
And until that happens meta studies are both worthwhile and one of the most effective ways to correlate and sensibly interpret the existing published data sets.
Also wouldn't junk itself have many subclassifications as how it is junk? Say poor study data quality, wrong assumptions of wider applicability, wrong analysis or conclusions, etc.