Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

On the whole, the Cochrane Collaboration is a force for good, but its methodology of reviewing previous published studies lets it miss important issues of prior plausibility (that is, lack of plausibility) when reviewing "alternative" treatments.

https://www.sciencebasedmedicine.org/of-sbm-and-ebm-redux-pa...




I don't think this has anything to do with Cochrane's methodology. This is about choice of presentation.

They are very clear in the paper that they are reviewing a certain body of studies, and that there are other plausible reasons (not backed up by RCTs) to believe water fluoridation could have beneficial effects. The numbers they quote summarize the evidential content of the RCTs. I don't need or want them to bundle this all into one Bayesian percentage. This would be dominated by their own priors which I do not necessary share nor are necessary shared by anyone I discuss the study with. (Yet we can all agree on the frequentist summary of the RCTs.)


Was that conclusion the result of a meta-analysis of meta-analysis methodologies? And there I thought this was just an amusing cartoon... https://xkcd.com/1447/




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

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

Search: