Hacker News new | past | comments | ask | show | jobs | submit login

> Is that "net" extracted from all published experiments, i.e. in some meta syntheses? And is statistically significance calculated in each individual experiment?

You can find excesses in both individual experiments and meta-analyses, IIRC.

> Publication bias toward positive results should be detectable via methods such as funnel plots.

Funnel plots have notoriously low statistical power, it would require only a little publication bias to produce excesses, and asymmetry in a funnel plot can be due to other reasons.




Thanks for these comments, will make for interesting investigations.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: