RWRMED: Stata module for performing causal mediation analysis using regression-with-residuals

rwrmed performs causal mediation analysis using regression-with-residuals. Using gsem, two models are estimated: a model for the mediator conditional on treatment and the pre-treatment covariates (if specified) after centering them around their sample means, and a model for the outcome conditional on treatment, the mediator, the pre-treatment covariates (if specified) after centering them around their sample means, and the post-treatment covariates (if specified) after centering them around their estimated conditional means given all prior variables. Thus rwrmed allows for the presence of treatment-induced confounders, which are post-treatment covariates that confound the mediator-outcome relationship.