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

Well, you can’t blind it but you can still randomize. If your outcome variables are sufficiently objective and important (e.g., mortality or cancer risk, not just perceived health), the possibility that the causality flows through a placebo effect is much less worrying. If [people correctly believing that they ate yogurt] reduces [the incidence of cancer], great!



I meant to say blind, not randomize, sorry.




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

Search: