3 ms·
Non-randomized sample and lots of reaching for stat-sig in the analysis. I gave up when they threw out the non-stat-sig result for their success criteria (29 da
by kleinsch 4y ago
Non-randomized sample and lots of reaching for stat-sig in the analysis. I gave up when they threw out the non-stat-sig result for their success criteria (29 days) and claimed victory bc 28 days was stat sig.
- srcreigh 4y agoThis is incorrect. The article had significant results for 2 groups (everybody and a subset of the participants) on the 29 day result, and also had significant result when looking at 28 days. They claimed the opposite - since multiple end points and groups all had significant weight loss, it’s not p hacking