| Debate about the role of scientific information in decision making processes has been around for millennia. Recently, this debate has been at the forefront in the context of the use of risk assessment for environmental, safety and health decisions. A troubling facet of the current debate is an entrenched dichotomy between two dominant philosophies---rationalism and social constructionism. Advocates on both sides focus on normative disputes over how society should use risk analysis, and they do so at the expense of a more fruitful consideration: how society can use risk analysis.; I contribute to this debate by analyzing the methods and data necessary for comparing diverse risks. I find that, even from a rationalist perspective, the resulting information may be poorly suited to prioritizing risk mitigation measures. I review the reference materials underlying a tabulation by the Harvard Center for Risk Analysis of "cost per life-year saved" estimates for five-hundred sixty life-saving interventions. I find that there are profound methodological inconsistencies across disciplines, such as bureaucratic analysts, medical epidemiologists and toxicologists. These differences are sufficient to account for at least some apparent differences in costs of various interventions. Similarly, by comparing the uncertainty surrounding a subset of those interventions, I find that the best available information only loosely supports prioritization among the interventions, and that error bars on some interventions span zero to infinity. I conclude that the decision to prioritize diverse risks along a single metric may undermine the social legitimacy of risk assessment in the policy process if it does not at least acknowledge technical shortcomings.; I conclude, however, that comparing risks is nonetheless a useful enterprise when we focus on two aspects of risks: disciplinary norms and uncertainty. While prioritizing diverse risks according to pre-determined criteria may be difficult or impossible, comparing inconsistent methods suggest opportunities to improve methods, and understanding the sources and nature of uncertainty suggests opportunities to reduce and manage uncertainty. I propose that if we institutionalize risk comparison at the federal level, we are best served by addressing methods and uncertainties, and not calculating end-points. |
