RT @JeremySussman: @ProfDFrancis @rwyeh @FelixValencia10 @willsuh76 @drjohnm To me, they estimate different things. ITT - effect of randomi…
@ProfDFrancis @rwyeh @FelixValencia10 @willsuh76 @drjohnm To me, they estimate different things. ITT - effect of randomization, similar to effect of MD prescribing a med. IV, what we call contamination-adjusted ITT, effect to patient of taking it. Both are
@JeremySussman you are now a made man. https://t.co/gr2ocolNth
The cited paper (https://t.co/t3esiEaJfN) makes an excellent case for instrumental variable analyses where MANY patients did not get their assigned treatment. For example, about half of patients assigned to screening don't show up for it.
@ProfDFrancis Could not tell from this thread if you are only arguing against as treated/per protocol analyses (agree, of course), or you also object to IV analyses being co-presented w/ unadjusted ITT (would be interested in why?). https://t.co/KGxojeT
@A_Schwa Or, you could use the contamination-adjusted ITT analysis: https://t.co/ZcRnZjHnN2
@myklebest We probably need to do this for our experiment. https://t.co/3n3dvQciPK
RT @metrics52: Keep trying! https://t.co/53Vlmieg7S
RT @JeremySussman: @rwyeh @MDaware @drjohnm @amitabhchandra2 @metrics52 Fwiw, we tried to explain this problem and the solution for docs a…
Keep trying! https://t.co/53Vlmieg7S
RT @JeremySussman: @rwyeh @MDaware @drjohnm @amitabhchandra2 @metrics52 Fwiw, we tried to explain this problem and the solution for docs a…
RT @JeremySussman: @rwyeh @MDaware @drjohnm @amitabhchandra2 @metrics52 Fwiw, we tried to explain this problem and the solution for docs a…
@rwyeh @MDaware @drjohnm @amitabhchandra2 @metrics52 Fwiw, we tried to explain this problem and the solution for docs a few years ago. https://t.co/2muzQbBXOA
@harlanyu @TheLab_DC In addition to nbreg, U really need to examine & report on stat power after adjustment 4 treatment contamination. https://t.co/KGxojeT1tL
@AgnesWold Här en artikel i BJM som diskuterar samma approach: https://t.co/wzrZvphaqc
RT @iwashyna: #PratoICM17, if worried re treatment contamination--as we are in these RBC RCTs, esp INFORM--maybe see https://t.co/Q8oefnrYpt
RT @iwashyna: #PratoICM17, if worried re treatment contamination--as we are in these RBC RCTs, esp INFORM--maybe see https://t.co/Q8oefnrYpt
#PratoICM17, if worried re treatment contamination--as we are in these RBC RCTs, esp INFORM--maybe see https://t.co/Q8oefnrYpt
RT @JeremySussman: @VinayPrasad82 @fgodlee we about the right way to address their problems in the BMJ, ironically. https://t.co/OBiXrMP8Ha
@VinayPrasad82 @fgodlee we about the right way to address their problems in the BMJ, ironically. https://t.co/OBiXrMP8Ha
Some options for analyzing experiments with non-compliance (from https://t.co/pbxDKAyVA2) https://t.co/AhoZosDuRW
@RogueRad @VinayPrasad82 @pash22 @davidcnorrismd Contamination bad. Of common solns, only ITT keeps randomization. https://t.co/bZlLa7ddBv
@RogueRad This limitation has partial solutions. Our intro: https://t.co/coWOjMxCoa
MT @angiologist IV = Instrumental Variable. A nice review from two colleagues at @UM_IHPI including @JeremySussman http://t.co/8FHyd9gTTG”
@angiologist IV = Instrumental Variable. A nice review from two colleagues at @UM_IHPI http://t.co/8FHyd9gTTG
How does intent to treat work? Pitfalls, limitations, other options? Two nice papers. http://t.co/fOZEUwhx6a http://t.co/EyP5htQKhP
How does intent to treat work? Pitfalls, limitations, other options? Two nice papers. http://t.co/fOZEUwhx6a http://t.co/EyP5htQKhP
IV comes in so many exotic names! Complier average causal effects, Efficacy estimator and Preference based analysis - http://t.co/aTY0yVzNu7
MT “@daviesbj Via @JeremySussman - how to read a contaminated RCT http://t.co/Q6thb7pBng” never knew - useful: most RCTs contaminated
Side product of off-line emails can be learning. Prof @JeremySussman alerts me to how to read a contaminated RCT http://t.co/iExYVVgVa0
Article on analyzing data from randomized experiment: http://t.co/QdoJkITwH9 #publichealth #research #RCT #medicine
Analyzing data from a Randomized Control Trial? Check out this article on how to analyze your data: http://t.co/QdoJkITwH9 #RCT #research