Model-Robust and Efficient Covariate Adjustment for Cluster-Randomized Experiments

Cluster-randomized experiments are increasingly used to evaluate interventions in routine practice conditions, and researchers often adopt model-based methods with covariate adjustment in the statistical analyses. However, the validity of model-based covariate adjustment remains unclear when the working models are misspecified, leading to ambiguity of estimands and risk of bias. In this article, we first adapt two model-based methods—generalized estimating equations and linear mixed models—with weighted g-computation to achieve robust inference for cluster-average and individual-average treatment effects. To further overcome the limitations of model-based covariate adjustment methods, we propose efficient estimators for each estimand that allow for flexible covariate adjustment and additionally address cluster size variation dependent on treatment assignment and other cluster characteristics. Such cluster size variations often occur post-randomization and, if ignored, can lead to bias of model-based estimators. For our proposed covariate-adjusted estimators, we prove that when the nuisance functions are consistently estimated by machine learning algorithms, the estimators are consistent, asymptotically normal, and efficient. When the nuisance functions are estimated via parametric working models, the estimators are triply-robust. Simulation studies and analyses of three real-world cluster-randomized experiments demonstrate that the proposed methods are superior to existing alternatives. Supplementary materials for this article are available online.