Investigating an Intervention’s Causal Story: Mediation Analysis Using a Factorial Experiment and Multiple Mediators

Abstract

Behavioral, biobehavioral, and biomedical interventions presume a causal story. This causal story is used to create a conceptual model of why the intervention caused the observed outcome. Methods are needed to investigate the extent to which changes in the outcome are due to exposure to an intervention. This chapter describes the use of a factorial experiment in which an intervention's components are varied, the mediating mechanisms and outcome are measured, and then an analysis is performed to investigate an intervention's causal story. To illustrate the procedures, a case study examines an intervention guided by the model of stigma communication (MSC; Smith, Commun. Theory 17:462-485, 2007; Stigma communication and health. In T. L. Thompson, R. Parrott, & J. Nussbaum (Eds.), Handbook of health communication (2nd ed., pp. 455-468). London, UK: Taylor & Francis, 2011; Commun. Monogr. 79:522-538, 2012). In the case study, we show how to conduct a mediation analysis when four components of an intervention have been manipulated in a 2 4 factorial experiment (N = 299), and, in addition, four hypothesized mediators have been measured. We reflect on how the results provide more precise conclusions about the theory guiding the intervention and opportunities for refinement.