Amazing, but this 1997 paper already shows what (and why) we can learn from observational studies about the bounds of treatment effects in a randomized experiment marred by non-compliance👇🏻The rest is just a history of the neglect by the EBM promoters. #causaltwitter

soboleffspaces's tweet image. Amazing, but this 1997 paper already shows what (and why) we can learn from observational studies about the bounds of treatment effects in a randomized experiment marred by non-compliance👇🏻The rest is just a history of the neglect by the EBM promoters. #causaltwitter

Glad you mentioned "non-compliance". Indeed, this was one of the first tasks addressed and solved in do-calculus. See ucla.in/2pQ0Gvr. The paper also includes applications to real-life data -- pacifying those who claim such applications are missing from CI literature.



Yes, it's a good paper. Alex Balke was one of my best students, and this paper was part of his PhD dissertation. Too bad he did not go to academia.


It continues to amaze me how much we had by 1997, when I started at Queen's, yet we still haven’t used in 2023. Have these 3 decades been spent on coming to terms (as if it wasn't known before) that EBM truthers and their truth container can't compete with the scientific method?


United States 趨勢
Loading...

Something went wrong.


Something went wrong.