Title: Using Soft Data to Make "Probabilisitc Risk Assessments" Realistic
Authors: Rex Brown, George Mason University
Date: April 1999
Status: working paper

A regulator or risk manager needs to assess the safety of an activity realistically, in a way that takes into logical account all knowledge that he believes to be relevant.  He will wish to build on all considerations, however inconclusive, that he already uses in his decisions (and does not wish to ignore).  He will also want to use any additional data he can develop or has access to, for example, using Probabilistic Risk Assessment.  However, PRA, as commonly understood, analyzes only those sources of risk on which there is hard data, and only takes hard data into account in assessing those risks.  Responsible executives are not satisfied to simply equate the uncertainty they will use in their decisions to a PRA, though they should certainly take that into account.  We propose a methodology, grounded in statistical decision theory, which accommodates soft as well as hard data.  It is illustrated in the context of nuclear and other environmental regulation.

Click here to access a pdf version of the paper

Back to the Decision Analysis Working Paper Index