Now that I’m going to Playroom regularly again, there was no escaping hearing about MRPW (multilevel regression and poststratification with weights), see the draft paper by Andrew Gelman, Yajuan Si, and Brady T. West. Here’s their setup: Suppose we have …