2 The term “mandated science” is taken from a recent, excellent book by Salter, Liora, Mandated Science (Dordrecht: Kluwer Academic Publishers, 1988)CrossRefGoogle Scholar.

3 The term “risk assessment” is generally used to refer to this first stage of risk analysis, so that the most fundamental distinction is that between risk assessment and risk management. Sometimes, risk assessment and risk estimation are distinguished: risk estimation is construed as one phase of risk assessment, the phase at which the level of risk or magnitude of risk is determined. Among those who make this distinction, there is little consensus concerning how the other phases of risk assessment should be identified. One example, though, would be hazard identification. We are focussing on risk estimation, determination of the level of risk. In this paper, the reader should regard the terms “risk estimation” and “risk assessment” as synonymous.

4 Of course the concept of “harm” is itself a normative one, but if it can be given a naturalistic definition (e.g., “incidence of cancer deaths in a population”), then the measuring of this harm is thought to be a purely empirical matter.

5 We mean by “normative” any judgment that reflects a political, moral, personal, or other value preference. These are generally recognizable by the presence of terms such as “good,” “better,” “ought,” “should,” and the like. Debates over these judgments generally are not settled solely by appeal to empirical evidence, since they are not debates about the facts alone, but also about the relative worth of various facts. For example, in the Alachlor debate the question, “How many farmers wore adequate protective gloves?” was primarily an empirical (non-normative) one. The question, “Should the exposure incurred by farmers who do not wear such gloves be included in estimates of applicators' exposure to Alachlor?” since it raised the issue of whether doing so would be fair to Monsanto, was a normative question.

6 Numerous authors have argued, as we do here, that risk assessment is not a value-free enterprise. Generally speaking, they fall into two groups. Some, looking to specific examples, represent the presence of values in the assessments as a sign that they are flawed. These critics, then, implicitly accept the ideal held forth by the classical view. They assume that value-free risk assessments are both possible and desirable; their contribution is to show that the ideal is seldom achieved. Others hold that the ideal is impossible to achieve, but base their argument on some general philosophical position, for example, the sociology of knowledge or a critique of the “fact-value” distinction. Our approach is similar to that taken by the first group (and unlike that taken by the second group) in that we do not argue from general principles but base our conclusion on an examination of particular cases. It is similar to that taken by the second group (and unlike that taken by the first group) in that we do not regard value-free risk assessments as possible. Nor do we regard assessments which fail to be value-free as necessarily flawed. For a view concerning risk assessment which is in many respects similar to our own, see Wynne, Brian, Risk Management and Hazardous Waste (Berlin: Springer-Verlag, 1987)CrossRefGoogle Scholar, and, with Aitken, L., “The System's Ability to Learn: Some Basic Problems in Post-hoc Accident Assessment,” in Policy Responses to Large Accidents, edited by Segerstahl, B. and Krömer, G. (Vienna: IFASA, 1989), p. 185–205Google Scholar. But see also Thompson, P., “Risk Subjectivism and Risk Objectivism: When are Risks Real?” Risk, 1 (1990)Google Scholar; Shrader-Frechette, K., “Scientific Method, Anti-Foundationalism and Public Decisionmaking,” Risk, 1 (1990)Google Scholar; McGarity, Thomas, “Substantive and Procedural Discretion in Administrative Resolution of Science Policy Questions: Regulating Carcinogens in EPA and OSHA,” Georgetown Law Review, 67 (1979)Google Scholar; Ashford, Nicolas et al. , “A Hard Look at Federal Regulation of Formaldehyde,” Harvard Environmental Law Review, 7 (1983)Google Scholar; Jasanoff, Sheila, “Contested Boundaries in Policy-Relevant Science,” Social Studies of Science, 17 (1987)CrossRefGoogle Scholar; Alvin Weinberg, “Science and its Limits: The Regulator's Dilemma,” Issues in Science and Technology (Fall 1985); Weinberg, Alvin, “Science and Trans-Science,” Minerva, 10 (1972)CrossRefGoogle Scholar; Jasanoff, Sheila, Risk Management and Political Culture (New York: Russel Sage, 1986)Google Scholar; Brickman, Ronald et al. , Controlling Chemicals (Ithaca, NY: Cornell University Press, 1985)Google Scholar; and Rescher, Nicholas, Risk: A Philosophical Introduction to the Theory of Risk Evaluation and Management (Washington, DC: University Press of America, 1983)Google Scholar. The Weinberg and McGarity papers place particular stress on the ways uncertainty in risk assessments leads to importation of assessors' value perspectives; in this way they anticipate what we shall refer to as “conditionally normative issues” in risk assessment.

7 The findings of the Review Board are printed in the Report of the Alachlor Review Board (Ottawa, ON: Alachlor Review Board, 1987)Google Scholar. The transcripts of the hearings are published as the Alachlor Review Board Hearings (Toronto, ON: Atchison & Denman Court Reporting Services, 1986, 1987)Google Scholar.

8 “Amortization” means simply the dividing of an exposure level by a fraction equal to that of the period of time out of a total lifetime during which a person is exposed. For example, if a person is exposed to 2.5 mg/kg/day for a total of 180 days out of an expected life-span of 72 years, the exposure of 2.5 would be reduced by a fraction of approximately 1/144. The assumption is that the carcinogenic effects of the chemical are not due to “one-shot” exposure, but are incremental with the length of exposure.