The headline is broadly defensible, but the qualifications matter. Effect sizes vary by population, the strongest claims rest on shorter trials, and credible voices push back on how it's typically framed.
Sayre we're going to do 10,000 subjects well you're clearly powered for it but you might be overpowered and people might say well why would that be a bad thing it could be a bad thing because it means you are very likely to reach statistical significance in things that might not be actually significant
Every Sunday: the week’s new conflicts and verdict changes — and nothing else.
Native comments, Twitter mentions, and Reddit threads about this claim — surfaced together so the conversation isn't fragmented across platforms.
Bookmarking — the dossier-vs-overview split is the right call. Most of the time I want overview; sometimes I want receipts.
Would love a "what would change this verdict" RSS feed. Sign me up if it exists.
what if you overpower a study in other words what if the power analysis says to be uh to have a level of power at 90% you need a thousand subjects and you Sayre we're going to do 10,000 subjects well you're clearly powered for it but you might be overpowered and people might say well why would that be a bad thing it could be a bad thing because it means you are very likely to reach statistical significance in things that might not be actually significant
big data are creating an opposite problem it means that you're overpowered and you can get statistically significant results that have no clinical meaning that have no meaning really and and even with a tiny little bit of bias you may get all these signals just because bias is there
i think that the growth of this type of problem will be faster compared to the growth of the problem of small underpowered studies