Often, in the case of experimental evaluations of multifaceted interventions, researchers and policymakers alike are interested in asking not only whether a given intervention had an effect but also why. They seek answers to questions such as: What features of the intervention led to the impacts that we have observed, or what was the causal mechanism or pathway through which random assignment to the intervention resulted in an improved outcome or set of outcomes? A variable on a causal pathway between an initial condition (such as treatment assignment) and an ultimate outcome is referred to typically as a "mediator." Conceptually, one regards the action of the initial treatment on the ultimate outcome as acting through the hypothesized mediator. Analytic strategies for assessing the extent to which the impact of an initial condition acts on the outcome in this way can be classified, broadly speaking, as approaches to mediation analysis. Quantitative methods for modeling mediational processes are an active area of exploration in the recent methodological literature. This paper contributes to the literature by illustrating an approach that capitalizes on the framework of principal stratification to address mediational questions in the context of an experiment. This use of principal stratification was proposed first by Rubin (2004) and has been applied and discussed recently by Gallop and colleagues (2009).