Chapter 8 Appendix: Additional topics of interest

The literature on mediation analysis has grown considerably in the last few decades and there are now many novel methods to tackle important questions with complex data structures. While we are unable to cover all these interesting methods in this workshop, here we provide a few references for further reading.

This list is not meant to be comprehensive, just some of our own work and some work by others that we know and consider interesting.

8.1 Mediation with multiple mediators and multiple intermediate confounders

Note that the medoutcon R package works for multiple mediators but only a single, binary intermediate confounder. If your data scenario includes multiple mediators and multiple intermediate confounders, you should use the HDmediation R package.

8.2 Mediation with time-varying treatments, mediators, and covariates

The issue of intermediate confounding is exacerbated in a setting with multiple treatments and mediators measured at various time points. Here are a couple of papers on this topic that are interesting:

8.4 Mediation with instrumental variables

8.5 Mediation with separable effects