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 ju

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!