The evidence ladder

Every claim on this site carries a level, and the level comes from the study’s design — never from how the paper describes itself. A preprint titled “definitive evidence” is still a preprint. The order below is the whole hierarchy.

Two rules do most of the work. The first: when signals conflict, the weaker level wins. A paper describing a rat model and a small human cohort is animal evidence, because promoting it on the strength of a mention is how overclaiming starts. The second: publication type beats study design. A review that discussesrandomised trials is not itself trial evidence — it is commentary about trials, and the honest sentence is “a review reported”, not “in controlled human research”.

Community discussion sits at the bottom and is never evidence at all. It tells you people are interested. That is a real signal about attention and a null signal about biology.