This talk describes recent advances in causal mediation analysis based on the potential outcomes (counterfactual) framework. Many fields apply mediating variable theory and traditional linear regression mediation analysis. The potential outcomes framework is a new and comprehensive approach to conceptualizing and testing questions about causal inference. I describe the motivation and general background for the potential outcomes framework for causal analysis. The potential outcomes framework is applied to mediation analysis, including the specification and identification of causal quantities and methods to estimate causal mediation effects. Similarities and differences between traditional and causal mediation analysis are highlighted. Both traditional and causal methods are applied to real data. The presentation ends with future directions for mediation theory and statistical analysis.
Topic Interest Group – Structural Equation Methods (SEM)
February 19, 2021 / 1:30 pm ET
- Panel Session
- Ask the Experts Session and Reception
- Login to your CARMA account.
- Once you login, in the middle, you will see an option to “Register for Live Events”. (If you have not registered your device yet, please register your device first. Then, click “User Area”.)
- Select the appropriate event and checkout. (Please note that separate registration is not needed for TIG meeting. Access info of both events will be shared in the same email.)
- You will receive an e-mail with the access information a few days before the event day.