Estimation of the Causal Mediation Analysis

Assuming identifiability, this chapter shifts focus to the specific estimation techniques in causal mediation analysis. It begins with familiar full parametric methods, but their shortcomings lead to the introduction of alternative methods, including weighting, imputation, nested effect, and simulation methods. The estimation of controlled direct effects and interventional (in)direct effects is also discussed. An illustrative example is provided, along with the replication codes.