Puzzles of Attributability

Summary

Measures of attributability are not mathematically complicated, but they are conceptually tricky. This chapter examines them with a view to achieving a clear and rigorous conceptualisation of attributable fractions. Two common errors in causally interpreting excess fractions are explored: the Exclusive Cause Fallacy, which supposes that the exposure causes all and only the cases represented by the excess fraction; and the Counterfactual Fallacy, according to which the excess fraction tells us how much the risk would drop by, absent the exposure. Instead we settle on an interpretation of “attributable to” as “explained by”, in line with the model of causal interpretation set out in Chapter 3.