ABSTRACT
Probability, causality and induction are central to clinical research. Causality has been given a lot of attention by philosophers and some statisticians in the last 100 years. This chapter examines Bradford Hill's consideration on causality. Michael Höfler has argued that Hill's conception of causality is best understood on a counterfactual account. Pearl examines causal modelling, central to which is his directed acyclic graphs (DAGs). DAGs are Bayesian networks (recall the discussion of Bayes' theorem in the previous chapter). The meaning of "graph" in this account is the one employed in the field of mathematics called graph theory. The philosophical debate about each of them continues to be engaged. Each has its defenders and critics. With attention on the purposes of seeking and employing causes in medicine (explanation, prediction and intervention), the complexity and variability of causes comes into sharper focus.
