The Generalized Oaxaca-Blinder Estimator

After performing a randomized experiment, researchers often use ordinary least-square (OLS) regression to adjust for baseline covariates when estimating the average treatment effect. It is widely known that resulting confidence interval valid even if linear model misspecified. In this article, we generalize conclusion covariate adjustment with nonlinear models. We introduce an intuitive way any “simple” construct covariate-adjusted The derives its validity from randomization alone, and models fit data better than models, it narrower usual OLS adjustment.