The Causal Interpretation Problem

Abstract

Why do epidemiologists worry and write about the nature of causation? This chapter focuses on one reason: the fact that measures of strength of association are also used as measures of causal strength. The extra import they have in this use is not captured by their mathematical definitions, presenting a problem of interpretation which even the most pragmatic epidemiologist must face when seeking to communicate her findings. Leading philosophical approaches to causation (probabilistic and counterfactual) do not provide useful solutions to this problem. The (novel) approach favoured instead asks whether the exposure explains the difference in outcome between the two groups, to at least the degree expressed by the measure of strength of association. A notion of singular causation is assumed, with the effect that general causal claims are reduced to quantitative claims about singular causation.