Causal Organic Direct and Indirect Effects: Closer to Baron and Kenny

• Activity Number: 45 - Statistical Models for Estimating and Testing Causal Effects in Biomedical Studies
• Type: Invited
• Date/Time: Sunday, July 29, 2018 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Judith Lok*
• Companies: Harvard T.H. Chan School of Public Health
• Keywords: Causal inference; Direct and indirect effect; HIV/AIDS; Mediation; Observational study; Organic direct and indirect effect

Abstract

Baron and Kenny (1986, 74,900 Google Scholar citations) proposed estimators of direct and indirect effects: the part of a treatment effect that is mediated by a covariate and the part that is not. Subsequent work on natural direct and indirect effects provides a formal causal interpretation. Natural direct and indirect effects use cross-worlds counterfactuals: outcomes under treatment with the mediator "set" to its value without treatment. Organic direct and indirect effects (Lok 2016) avoid cross-worlds counterfactuals, using "organic" interventions on the mediator while keeping the initial treatment fixed at "treatment". They apply also to settings where the mediator cannot be "set". If there is no treatment-mediator interaction, both natural and organic indirect effects lead to the same estimators as in Baron and Kenny. In this talk, I propose organic interventions on the mediator while keeping the initial treatment fixed at "no treatment", leading to an alternative version of organic direct and indirect effects. I will show that the product method, proposed in Baron and Kenny, works for this indirect effect even if there is treatment-mediator interaction. Furthermore, I will argue that this alternative organic indirect effect is more relevant for drug development than the traditional natural or organic indirect effect.