Causal effects in mediation analysiswith limited-dependent variables

Mediation is used to separate direct and indirect effects of an exposure variable on anoutcome variable. In this thesis, a mediation model is extended to account for censoredmediator and outcome variable. The two-part framework is used to account for thecensoring. The counterfactual based causal effects of this model are derived. A MonteCarlo study is performed to evaluate the behaviour of the causal effects accounting forcensoring, together with a comparison with methods for estimating the causal effectswithout accounting for censoring. The results of the Monte Carlo study show that theeffects accounting for censoring have substantially smaller bias when censoring is present.The proposed effects also seem to have a low cost with unbiased estimates for samplesizes as small as 100 for the two-part mediator model. In the case of limited mediatorand outcome, sample sizes larger than 300 is required for reliable improvements. A smallsensitivity analysis stresses the need of further development of the two-part models