Extending the Baron and Kenny Mediation Analysis to Allow for Exposure-Mediator Interactions: SAS and SPSS Macros

• Activity Number: 415
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section
• Author(s): Linda Valeri*+ and Tyler J. VanderWeele
• Companies: Harvard University and Harvard University
• Keywords: Mediation ; Interaction ; Software

Abstract

Mediation analysis is a useful and widely employed approach to studies in the field of psychology and in the social and biomedical sciences. The main contribution of the present work is providing SAS and SPSS macros for easy implementation of mediation analysis allowing for the presence of exposure-mediator interactions. The software enables an investigator to carry out mediation analysis under a regression framework, building on the traditional approach proposed by Baron and Kenny (1986). This traditional approach does not allow for the presence of exposure-mediator interaction in the mediation analysis model and our aim is to fill this gap. We describe identifiability conditions and counterfactual definitions for the causal effects estimated via mediation analysis with a particular emphasis on the case in which exposure-mediator interaction is present. This is accomplished by applying and extending the work on identification and estimation of direct and indirect causal effects of VanderWeele and Vansteelandt (2009,2010).