Kevin Lewis points to a research paper and writes, “I’m not sure how robust this is with just some generic survey controls. I’d like to see more of an exogenous assignment.” I replied: Nothing wrong with sharing such observational patterns. … Continue reading