2,426 followers
@statsepi @stephenjwild @SolomonKurz @noah_greifer @tmorris_mrc this paper looks promising and specifically calls out RCTs as a situation where missing indicators are valid: https://t.co/4jxB00vVAg
12,367 followers
@chrisbboyer Interesting and good to see this message disseminated. Also agrees with earlier work on this topic including https://t.co/AsLqcs5wzr and https://t.co/lKqWugURla
3,170 followers
@MaartenvSmeden @stats_q @_MiguelHernan @JAMAInternalMed @salimhayek @ajadmon @HarvardChanSPH @BrighamWomens @tmorris_mrc @IanWhit25399993 @t_mpham Thanks Maarten. https://t.co/EgNZhlsMZv explains why for confounder adjustment it is not generally a good ap
10,492 followers
@ShalitUri @kdpsinghlab @ampanmdagaba @HerrDoktorFunk @venkmurthy @MaartenvSmeden @Richard_D_Riley @ADAlthousePhD @f2harrell @ESteyerberg @NavTangri How is 'works as well' defined? Missing indicator is often quite a poor approach with unpredictable bias (h