All simulators
Double Robustness (AIPW)
Switch the outcome and propensity models between right and wrong: AIPW survives as long as one is right.
Day 45: Phase 3 Capstone: Doubly Robust (AIPW) EstimationOutcome modelPropensity model
Regression imputation — misspecified
bias 0.624
sd 0.158
IPW (Hájek) — should be consistent
bias 0.020
sd 0.189
AIPW (doubly robust) — should be consistent
bias 0.019
sd 0.174
Regression imputationIPW (Hájek)AIPW (doubly robust)
Workplace wellness programme; outcome: days of work lost. Both the outcome and the enrolment depend on BMI non-linearly, so a model with only a linear term is wrong. AIPW is consistent if either the outcome model or the propensity model is right; it fails only if both are wrong. True ATE = 2.