Battle in the planning office: Biased experts versus normative statisticians

For the purposes of calculation, context is irrelevant: one is expected to strip away the "contingent" details, slot bare numbers into the equations, and perform the relevant maths. Medical doctors must know this. So why, asked just such a question about the likelihood of a diagnostic test being accurate, do two thirds of respondents get the answer wrong? These results are usually used to demonstrate the medics' woeful comprehension of probability theory. This paper, however, argues that the results can be understood as a reminder of the importance of context to the constitution of "rationality". Reinterpreting the results in light of "ecological rationality" - which takes account of context - reveals that the problem may not be with the respondents, but with the conception of rationality as necessarily context independent. "Facts" are statements about the world for which there is consensus, and consensus will be achieved when a statement can be accepted on rational arguments. But what kind of arguments can be considered as rational?