Several multiattribute scoring systems have been developed in response to the need for government agencies to set priorities for testing chemicals for toxic effects, including carcinogenicity. The agencies directly involved have been the National Toxicology Program in the Department of Health and Human Services, and the Environmental Protection Agency (EPA), which is required under the Toxic Substances Control Act to publish every six months a list of the highest priority chemicals it has identified for mandatory testing in the private sector.
A committee of the National Research Council/National Academy of Sciences reviewed a number of the scoring systems that had been developed (National Academy of Sciences, 1981). One of the more sophisticated systems developed for the Environmental Protection Agency is described in the report of the Toxic Substances Control Act Interagency Testing Committee (1977). This scoring system is based on two groups of attributes: exposure-related and biological. The exposure score used in this method consists of a weighted combination of four factors: quantity of production; quantity released; number of persons occupationally exposed; and number of persons generally exposed. The “quantity released” factor score is a composite of two subfactor scores (quantity and persistence), and the factor score for “extent to which the general population is exposed” is a composite of four subfactor scores (number of people exposed, frequency of exposure, exposure intensity, and penetrability). Interestingly, although the latter four subfactor scores are added, they are quantified on a quasi-logarithmic scale so that the process approximates a multiplicative combination, which would be the approach dictated by decision analysis.
The biological scores used in the method seem more arbitrary and are actually used only informally (i.e., in a multiattribute accounting mode) in the selection process. Ultimately, a list of the few dozen top candidates, as identified by the scoring system, is reviewed by the committee, and the final selection is based on informal judgment, not the formal scores. This subjective final step is necessary because no attempt is made earlier to incorporate within the scores several key objectives, including feasibility of testing and regulation.
Cost-effectiveness analysis also has been suggested for the testing priorities problem (Weinstein, 1979), and multiattribute scoring systems have been proposed for setting priorities for regulating toxic chemicals (Squire, 1981).
A recent example of a multiattribute scoring system for project selection has been developed at the EPA in response to legislation that sets aside a certain amount of money for the clean-up of the highest priority hazardous waste sites in the nation—the so-called Superfund. The scoring system, called the Hazard Ranking System (HRS), is part of