Causal parameters without corresponding experiments: powerful abstractions or pitiful distractions?

Causal parameters without corresponding experiments: powerful abstractions or pitiful distractions?

The objective of Epidemiology is to help improve population health. To do so requires the development and evaluation of interventions based on the causal determinants of health outcomes. Although there have been great advances in causal inference, areas of significant debate remain. In this symposium, speakers will challenge each other on three of the most controversial topics. The first debate “Manipulation is not required for causation” will discuss estimating the effects of variables that do not correspond with well-defined interventions, such as obesity, sex, and race. The second debate focuses on methods to evaluate mediation: “Natural effects are natural and meaningful” will highlight the differences and limitations in interpretations of natural and controlled direct effects. The third debate will discuss the advantages of principle stratification and whether its underlying assumptions are plausible or useful in health research.

Session Chair:
Ian Shrier, McGill University

Introduction:
Timothy L. Lash

Editor-in-Chief, EPIDEMIOLOGY
Past President, SER

Presenters:
Maria Glymour, University of California, San Francisco
Miguel Hernan, Harvard T.H. Chan School of Public Health
Ian Shrier, McGill University
Richard MacLehose, University of Minnesota
Bryan Shepherd, Vanderbilt University
Sonja A. Swanson, Erasmus Medical Center

Moderator:
Jay Kaufman, McGill University