Kim Kaivanto & Winston Kwon
In this paper we attempt to recover an integrated conception of the Precautionary Principle (PP). The alpha=0.05 inferential-threshold convention widely employed in science is ill-suited to the requirements of policy decision making because it is fixed and unresponsive to the cost tradeoffs that are the defining concern of policy decision making. Statistical decision theory — particularly in its Signal-Detection Theory (SDT) variant — provides a standard framework within which to incorporate the (mis)classification costs associated with deciding between preventive intervention and non-intervention. Circumstances that yield a (1,1) corner solution in the SDT model are also sufficient for preventive intervention under the PP. Thus the PP can be understood as a heuristic variant of the SDT corner solution, where the SDT model serves to patch the incongruity between the inferential practices of science and the inferential requirements of policy decision making. Furthermore, SDT’s analytical structure directs attention to a small number of variables — (mis)classification costs and prior probabilities — as determinants of the (1,1) corner solution. Psychological biases impinging upon these variables — omission bias and protected values — combine within SDT to successfully retrodict features of the PP previously considered puzzling, if not inconsistent or incoherent.