In this commentary, we discuss one potential way to further improve the ratio-of-mediator-probability weighted (RMPW) estimation methodology for causal mediation analysis. Although the RMPW methodology does not involve explicit modeling of the mediator-outcome relationship, it still requires researchers to correctly model the mediator given the treatment and pretreatment confounders—a challenging task in the presence of many confounders. We propose the use of the covariate balancing propensity score (CBPS) estimation as a more robust alternative that exploits the dual characteristics of the propensity score to improve the estimation of causal mediation effects.

mediation: R Package for Causal Mediation Analysis — CRAN / GitHub