Developments in the last thirty years have led some authors to formulate the basic problem of statistics as being rational decision making in the face of uncertainty. This formulation is both suggestive and unclear. On the one hand, it suggests that modern statistical theory can treat from a single viewpoint problems where one must decide how to act in the face of uncertainty and problems where one must decide what to believe in the face of uncertainty. On the other hand, this formulation raises the question as to whether the unity of viewpoint is due to certain structural similarities between different kinds of problems or whether it is based on a debatable reduction of problems concerning what to believe to practical problems.

Clarification of this last point is of considerable philosophical interest. Applications of statistical procedures are to be found in physics, genetics and the social sciences in contexts where the major objective of inquiry seems to be to accept true propositions as true (i.e., where the problem is to decide what to believe). Statistical procedures are also used in determining insurance rates, in market research, quality control, etc. where the problem is to decide how to act. The fact that similar procedures of inference can be applied to theoretical and practical problems has important implications for any conception of the relations between the theoretical and the technological and policy making aspects of scientific activity. The issue at stake stems from the fact that current statistical theory justifies its procedures by appealing to practical applications of these procedures.