[gwern]

> It’s important to note: the study only had 20 participants, so it wouldn’t have enough statistical power to detect a moderately large effect.

This is a within-subject experiment: each participant got all 4 possible doses. The paper (https://www.gwern.net/docs/nootropics/2019-bershad.pdf) doesn't discuss power that I can see, but the power of a within-subject design will be a lot higher than the between-subject you're probably thinking of, where n=20 is generally way too little. So, might be decently powered after all.

[belltaco]

So the article did not do a good job of explaining the study.

Also, why did they not go for a dose every 3-5 days like the anecdotes, but instead of went for every 7 days. And then say it might be different when done every few days. Were they trying to avoid testing on the weekends or something?

[gwern]

Washout. If you went more frequently, then the excuse would be that the LSD microdosing might still be having effects on the low-dose or placebo days.