Doubly robust estimation

Doubly robust estimation combines two nuisance-function estimates so that the target estimator remains consistent when either one of two model components is correctly specified.

This estimator combines outcome regression with inverse propensity weighting and has a second-order product bias.

An estimator is doubly robust when it remains consistent if either of two nuisance models is correctly specified, commonly an outcome-regression model or a propensity-score model.

Cross-fitting trains nuisance estimators outside each observation's fold and evaluates the target score on the held-out fold.

Double machine learning uses orthogonal estimating equations and cross-fitting so that sufficiently accurate machine-learning nuisance estimates yield asymptotically normal target estimates.