MRStdCRT
Implements model-robust standardization for cluster-randomized trials (CRTs). Provides functions that standardize user-specified regression models to estimate marginal treatment effects. The targets include the cluster-average and individual-average treatment effects, with utilities for variance estimation and example simulation datasets. Methods are described in Li, Tong, Fang, Cheng, Kahan, and Wang (2025) <doi:10.1002/sim.70270>. Inverse-cluster-size weighted outcome models are available for generalized estimating equations and linear or generalized linear mixed models. The outcome models can use common fitting weights for both estimands or inverse-cluster-size weights for the cluster-average effect and no weights for the individual-average effect. Inference retains the original standardization framework and paired cluster jackknife. Related cluster-size saturated models are described in the arXiv preprint by Lee, Harhay, and Li (2026) <doi:10.48550/arXiv.2608.28943>.
This page was built for software: MRStdCRT
