Statisticians have a rule of thumb for calibrating claims made in humanities and science papers alike. Andrew Gelman, for example, talks about statistical significance filter
- "If an estimate is statistically significant, it’s probably an overestimate."
A good thing to remember when you read weak observational studies, psychology surveys and, in modern times, a shocking number of epidemiology papers.
For health, you can use a different rule of thumb: Does Joe Mercola sell it?
If he does, it is probably suspect.