Mixed methods mediation analysis: Method and application to a tobacco prevention program evaluation

Mediation analysis investigates a hypothesized causal chain between independent, dependent, and mediating variables. While methods for conducting statistical mediation analysis have been studied at great length, mediation analysis using qualitative methods has received little attention. Qualitative research offers opportunity to make causal explanation by observing and recounting causal mechanisms and the contexts within which they operate. This paper introduces a mixed methods approach to mediation analysis and applies the proposed method to an evaluation of a tobacco-use prevention program, Project MYTRI (Mobilizing Youth for Tobacco-Related Initiatives in India), for adolescents in India. The proposed method employs (a) quantitative investigation to identify and test causal effects and (b) qualitative investigation to explain causal processes and contextual factors. The application demonstrates how a mixed methods mediation design can result in an enriched investigation of causal mechanisms beyond that which could be achieved through statistical mediation analysis alone.