(2 papers I like on this topic: https://t.co/yA4IA5upLO https://t.co/stX8iQkC3x)
375 followers
2,584 followers
@PaulKirvan @je_ff_ff @aca_medic @VPrasadMDMPH Actually, the main result had a P value ~ 0.05, and they did not do an "as treated" analysis, but did an instrumental variable analysis. Here's a paper I wrote on this approach. https://t.co/pvAhspblQJ
2,157 followers
@instrumenthull @metrics52 I've been teaching this to doctors for many years. The key reason it hasn't stuck is that in medicine effect sizes just aren't that important. The Q's want to make negative trials positive (nope!), but making small effects a bit
2,584 followers