Evaluating the possible application of Probabilistic Risk Assessment (PRA) to engineering ethics, Paul B. Thompson of Texas A&M University cautioned that ultimately PRA "ought to be used to make technologies safer; it ought not to be used as a proof that a technology is already safe." Though PRA can provide a model of a technological system which provides "a mathematical technique for estimating the probability of events which entail physical damage or loss of life," he maintained that instituting PRA does not discharge engineers and planners from their ethical responsibilities. Furthermore, though Thompson contended that PRA can contribute to the field of public policy analysis, he admitted that its implementation cannot and should not constitute a basis for ethical decision making.

Since the reactor Safety Study of the seventies was the first PRA ever attempted, Thompson used it as an example for surveying PRA techniques . . . He claimed that PRA serves two functions: 1) to provide "some sort of model of plant systems which, when quantified, yields an estimate of the likelihood of accidents," and 2) to provide "an estimate of the consequences ensuing from a given type of accident.

When applied to a technological system, both analysis techniques are multiplied; this results in a single number which represents ' the risk. ' However, Thompson noted that," whether or not such a number truly represents the risk as studied by PRA is yet another point of contention among risk analysts."

Thompson maintained that increasing technical complexity as well as economic factors have made it "more difficult for an individual engineer to bear in mind external factors in interactions which jeopardize safety and reliability." In this regard, he sees the contribution of PRA as an indicator of whether a situation merits further attention rather than as a basis for public policy decisions. He believes that implementing PRA as such would reduce instances of "whistle-blowing" because "the engineer's fulfillment of ethical responsibilities would become more clearly consistent with management's desire to protect capital. Ethics would cease to be a negative drain on capital."

Finally, Thompson reminded his audience that without an objective commitment on the part of its supporters to PRA as an independent, ongoing process, its value is negligible. He concluded with the following observation: "It is clear that the ethical dimension of PRA depends upon the ability of analysts and the model to adjudicate controversial claims in a fair manner. If the project feels pressured to skew results in order to favor one view rather than another, the potential benefits of PRA are lost."

Professor Edwin Levy of the University of British Columbia and Professor David Copp of Simon Fraser University pointed out several ways in which ethical and value considerations enter into risk assessments and cost-benefit analyses. In regard to the former, they noted the following distinction between two kinds of comparisons which each play crucial roles in arguments about the acceptability of risks. The first kind, referred to by Levy and Copp as "legitimating" comparisons, are introduced specifically to argue for the acceptability of a given risk. A legitimating comparison draws an analogy between the risks associated with a controversial and, hence, acceptable nature. To the second type of comparison Levy and Copp appended the designation "metrostick." Metrestick comparisons are not intended to convince one of the acceptability of a given risk. Instead, they serve to make its magnitude comprehensible to the general public, which is unfamiliar with the phenomena, terminology, and statistics at issue.

Levy and Copp pointed out that "for [a] legitimating comparison to be successful it must be shown the ethical features of the situations being compared are on a par. For example, it must be shown that the data involved are equally reliable and that the situations of those at risk are approximately the same with respect to involuntariness." By contrast, they noted, "[in] a metrestick comparison one need not demonstrate the ethical parity of the parties at risk."

Levy and Copp called attention to several misuses of legitimating and metrestick comparisons, which they consider widespread. In some instances, they said, "metrestick comparisons play the rhetorical role of legitimating comparisons; and the weaknesses of the legitimating comparisons which are offered are less discernable in the muddied waters." Levy and Copp also asserted that "legitimating comparisons have beguiling but insupportable implications: they suggest that what is already accepted is acceptable; that is, just because a practice is accepted certainly does not show that that practice is-morally or otherwise-acceptable. Furthermore, especially in the domain of public policy, just because a particular practice is followed does not necessarily show that it is accepted."

In the area of cost-benefit analysis, Levy and Copp decried the practice of "(taking) a mix of moral viewpoints into account within a cost benefit orientation . . . by counting a person's moral disapproval of a proposal as a 'psychic' cost and moral approval as a'psychic' benefit." Such an approach, they complained, "treats ethically based disapproval [as] data to be tabulated rather than a substantive issue to be addressed. It fails to treat members of the community as peers of the decision maker."

Levy and Copp concluded by saying that "it is unavoidable and desirable that ethical considerations be brought into play in policy analysis. But it is essential that this be done overtly so that the role which ethical considerations play can readily be discerned.

Commentator Professor Michael Martin of Chapman College suggested that the considerations Levy and Copp treat as constraints upon risk assessment and cost benefit analysis might be directly integrated into those procedures. Martin suggested that the notion of a rights based as well as a consequence based cost-benefit analysis deserves further exploration. With respect to Thompson's paper, Martin stated that in his opinion, "Thompson greatly overestimates the benefits of [probabilistic risk assessment] for ethical decision making for engineers and others within a corporation." Martin contended that the ethical standards of corporate decisions are affected by a variety of diverse factors which at best, will only be marginally affected by the adopting of probabilistic risk assessment techniques. The impact of such techniques, he said, will vary enormously with differences of organizational structure, particularly in regard to the degree of decision making discretion granted engineers.