When you read a paper with a DiD design & observational data, do you consider it causally-identified?
No, causality is only with RCTs 17.6% (21)
Only in rare cases (natural experiments) 28.6% (34)
Sometimes (depends on # of tests) 28.6% (34)
Usually (DiD works well IMO) 25.2% (30)
119 interactions
Posted by @rmkubinec.bsky.social on 4/9/2026, 3:31:55 PM