Recent developments within causal inference based effect separation (ie. mediation)
主 题: Recent developments within causal inference based effect separation (ie. mediation)
报告人: Dr. Theis Lange, Associate Professor in biostatistics (University of Copenhagen)
时 间: 2015-12-10 14:00-15:00
地 点: 理科一号楼 1114
Within the fields of epidemiology, interventions research and social sciences researchers are often faced with the challenge of decomposing the effect of an exposure into different causal pathways working through defined mediator variables. The goal of such analyses is often to understand the mechanisms of the system or to suggest possible interventions. In this talk I present the class of natural effects models, which allows causal inference based mediation analysis to be understood (and performed) in terms of traditional regression analyses. I will introduce the model class and the interpretation of the parameters in the model. Next I will present a just published R-package, which makes estimation of natural effects models both simple and intuitive. Finally, I explain how natural effects models can be used to address more complex mediation questions; in particular multiple mediators. The talk will include recent work on mediation analyses of a randomized clinical studies (RCTs) and epidemiological studies.
