We read with interest Jay Kaufman's1 commentary on our paper, 'Opening the black box: a motivation for the assessment of mediation',2 in which he takes us to task for focusing on natural effects (i.e. description of how the exposure influenced the disease through the mediator), rather than controlled effects (i.e. estimates of the effect of intervention on the mediator). Kaufman's critique raises an important problem that goes well beyond mediation-the relationship between aetiological research and the goal of improving the public health. We welcome this opportunity to clarify our perspective.We take as a starting point, as does Kaufman, a counterfactual approach to causal questions. From this perspective, our analyses are based on a thought experiment that uses observed substitutes for the desired but unobservable comparisons of causal questions.3 The appropriate thought experiment depends on the goal of the study.
展开▼
