ABSTRACT
This chapter explores the roles and uses of science data and theory within the field of risk analysis, in hopes of tying the two more securely together in current conception and future practice. It shows that the rationality of employing past data to make predictions of future risk under potentially changing antecedent conditions requires the explicit use of causal theories. These theories are the means by which rational confidence is formed through evidential reasoning. The chapter presents the example of ranking risks from environmental exposure to carcinogens. It looks at the opening the possibly controversial claim that tables listing expectation values of the fraction of people dying after environmental exposure to various carcinogens do not necessarily provide a rational basis for making public policy decisions or even for summarizing the state of scientific understanding. The chapter expresses that a particular estimate of attribute frequency carries an epistemic status related to the degree of crafting represented by that estimate.
