A Practitioner’S Guide To Intent-To-Treat Effects From Multisite (Blocked)...

on YouTube ↗27m

" summary

Defines four concrete estimands (finite or super population crossed with person average vs site average) and catalogs about 15 estimators applied to 12 real multisite trials and simulations. Finds differences in point estimates usually tiny but sometimes up to ~0.1 SD, and standard errors can differ two to fourfold with super-population SEs unstable and hard to estimate. Covariate adjustment rarely cuts SEs by more than ~10%. Provides an R package that runs all estimators and argues analysts must state the estimand and SE framework up front.

AI-written summary · about

" chapters (5)

1. 0:00Start
2. 16:45Does choice of estimand/estimator matter for the estimate of B?
3. 18:34Does choice of estimand/estimator matter for the estimate of SECB ?
4. 22:20Bias Precision Trade-off?
5. 24:16And what about covariate adjustment?

" related talks

• 01h 11mStatistical Limits of Causal InferenceSivaraman Balakrishnan · Modern Paradigms in Generalization Boot Camp 2024
• 33mSimple Yet Efficient Estimators For Network Causal Inference...Christina Yu · Algorithmic Aspects of Causal Inference 2022
• 45mInference and Interference in Marketplace ExperimentationHannah Li · Quantifying Uncertainty: Stochastic, Adversarial, and Beyond
• 54mDebiasing Random Forests For Treatment Effect EstimationJasjeet Sekhon · Statistics in the Big Data Era 2022
• 35mEstimating the Value of PersonalizationEmma Brunskill · Bridging Prediction and Intervention Problems in Social Systems (Simons Institute, 2026)
• 39mCausal Effect Estimation in Practice: Lessons Learned from E-commerce & BankingDanial Senejohnny · PyData Amsterdam 2024
• 27mDreadful Frailties in Propensity Score Matching and How to Fix ThemAlexandre Abraham · PyData Paris 2024
• 17mMaking sense of marginal effectsDemetri Pananos · posit::conf(2024)