Cover Image

PAPERBACK
$30.75



View/Hide Left Panel
Click for next page ( 2


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 1
PERSPECTIVES ON RISK AND DECISION MAKING When people say they must make a decision involving risk, they often mean that the decision involves the possibility of an adverse consequence. This report does not deal with risk and decision making so broadly de- fined, since almost every important decision in life entails the possibility of an adverse consequence. To narrow its task, the committee concentrated on those decisions involving the possibility of adverse effects for human health, safety, and the environment. This narrowing does not reflect a belief that other types of risks are less important--certainly the personal risks in a career choice or the economic risks faced by people in business are no less important. The focus reflects the committee's expertise and, more important, a growing public awareness of and concern about adverse conse- quences to human health, safety, and the environment. INCREASED LONGEVITY AND RISING PUBLIC CONCERN ABOUT RISK The increases in average life expectancies for Americans have not lessened their concern with risks and may even have increased it. The evidence for changing mortality and morbidity is worth examining briefly as a base for

OCR for page 1
2 gauging actual risks; further exploration of possible reasons why concerns with risk have not abated may im- prove understanding of the perceptions of risk. The first sentence of a recent report (U.S. Surgeon General, 1979) proclaimed that "the health of the Ameri can people has never been better." ~' -Lee surgeon general documents improvement in many indices of illness, non- fatal conditions, and mortality. We would like to emphasize primarily the progress made against mortality, since, in the absence of a comprehensive statistical measure of risk, the single best indicator is perhaps that provided by mortality statistics. The decline in U.S. mortality rates since the turn of the century has been steady and, on a cumulative basis, dramatic. Life expectancy at birth has increased in the United States from 47 years in 1900 to 74 years in 1979. The age-adjusted death rate has fallen by two-thirds, from 18 per 1,000 in 1900 to under 6 per 1,000 in 1979. The probability of "early" death--death before age 65--has declined from over 60 percent at 1900 mortality rates to under 25 percent at 1979 rates. Mortality reductions in the 1970s have been espe- cially impressive. For example, in the United States the likelihood that a person age 65 would live at least another decade has increased by 14 percentage points from 1900 to 1970 (from 55 to 69 percent) and then by another 5 percentage points by 1979 (to 74 percent). In 1979, the age-adjusted death rate at all ages was 18 percent lower than it was 9 years earlier. Similar im- provements have occurred in other countries as well, including dramatic ones in the less developed countries considered as a group. According to one poll (Louis Barris and Associates, 1980) most Americans believe life is getting riskier: 78 percent of the public surveyed agreed that "people are subject to more risk today than they were twenty years ago" (p. 9~; only 6 percent thought there was less risk. Furthermore, 55 percent felt that "the risks to society stemming from various scientific and technologi- cal advances will be somewhat greater 20 years from now than they are today" (p. 10), as opposed to only 18 per- cent who felt that the risks will be somewhat less. The degree of concern about risk has increased sharply since about 1960. For example:

OCR for page 1
3 . . . . . . . The enactment of more than 30 major laws by the U.S. Congress from 1965 to 1980 (and numerous laws by state legislatures) aimed at coping with occupational, consumer product, environ mental, transportation, and other sources of risks; The establishment or strengthening of at least a dozen regulatory agencies with broad legal authority and rising budgets throughout the 1970s; The growth of litigation related to health, safety, and environmental risks in the 1970s in both the tort-liability system and the arena of judicial review of agency decision making; The creation and growth of numerous public interest groups concerned with health, safety, and environmental risks, a movement that has significantly changed the politics of risk and the politics of regulation; The emergence of various forms of business sponsored efforts to improve risk management by the private sector, to publicize these efforts, and to coordinate them with those being taken by local, state, and federal governments (for examples of increasing corporate efforts to control risks, see U.S. Department of Commerce, 1980~; The expanded media coverage of scientific find ings about risks, corporate risk .anagement activities, and political activity related to risk; m e increase in funding for health, safety, and environmental research and the emergence of a relatively new field of "risk analysis." SOME POSSIBLE INTERPRETATIONS Multiple hypotheses have been proposed for the apparent contradiction between increased longevity and increased concern with risk.1 While several hypotheses are given below, there are undoubted!, others. . iSeveral reviewers and committee members comment that they feel no compelling reason to explain "the apparent contradiction between increased longevity and increased

OCR for page 1
4 Mortality Differentials One source of concern about risk may be a growing reali- zation that certain groups in society suffer from higher than ordinary rates of early death. Despite historical progress, it is increasingly apparent that there are numerous opportunities for further longevity gains. For example, two types of mortality differentials suggest that it may still be possible to substantially reduce mortality rates in the United States: the high rates of mortality among disadvantaged groups in the United States and the higher rates of mortality for the entire U.S. population compared with that of many other developed nations. Blacks, Hispanic Americans, American Indians, and the poor and poorly educated in general suffer from sub- stantially higher death rates than do middle- and upper- class whites. mese disparities should be understood in the context of a period of dramatic improvement, at least for some disadvantaged groups. Between 1950 and 1975, for example, both whites and nonwhites experienced sharply reduced rates of infant mortality as well as general increases for all ranges of age-specific mortal- ity except for the very old. Although these improve- ments were greater for nonwhites than for whites, mor- tality differentials between the two groups remain large. For example, if current mortality rates remain unchanged, two of five nonwhites will die before reach- ing age 6S, compared with one of five whites. Although data are sparse, mortality and morbidity rates among migrant workers in the United States appear to be espe- cially high. The United States fares poorly compared with other developed countries in terms of a variety of health in- dicators, including infant mortality, life expectancy, cardiovascular disease rates, cancer death rates, and homicide rates. For instance, the United States is 26th on a list of countries in the probability of death be- fore age 6S. In terms of reported mortality rates, the concern about risker They go on to say that in their opinion there i8 no real contradiction because of reasons A, B. or C. Not all these critics, however, agree on A, B. or C. "An apparent contradiction" is an individual opinion: some believe it; others don't. - , v ~ . .

OCR for page 1
United States ranks close to countries with less than half the U.S. per-capita income. The likelihood of death before age 65 for selected countries is: Rank Country Probability 1 Sweden .183 10 Canada .235 26 United States .275 37 Mexico .419 41 Liberia .665 U.S. whites, however, have a probability of death before age 65 of .412. Persistent differences in life expectancies, between nations and between different groups in the United States (e.g., whites and nonwhites), may mean that, for the lagging populations, achievable mortality gains are not being realized. These attainable gains also imply inequities in the distribution of risk: some croups carry higher burdens of risks. ~ r ~ tality differentials and the linked issue of equity in the sharing of risks are surely part of the reason for increased concern with risk despite general increases in life expectancies. Smoking and Drinking The persistence of mor Some mortality gains are attainable through individual action. For example, alcohol consumption ranks second to cigarette smoking as a behavioral cause of death. The report by the U.S. surgeon general (1979) estimates that "alcohol misuse is a factor in more than 10 percent of all deaths in the tinited States" and cigarette smok- ing is a factor in 17 percent of all deaths. Nonethe- less, alcohol abuse produces social problems that, in at least two respects, are more serious than those problems caused by cigarette smoking. First, alcohol abusers impose substantial costs on others (especially on family and close friends) through homicide, suicide, spouse beating, child abuse, and motor vehicle accidents. . . . . Second, alcohol abuse is a major cause of death at younger ages. For instance, the surgeon general reports that "alcohol-related accidents are the leading cause of death in the 15 to 24 age group."

OCR for page 1
6 Cigarette smoking remains a major health problem, despite reductions in the numbers of Americans who smoke. As the surgeon general's report emphasizes: Cigarette smoking is clearly the largest single preventable cause of illness and pre- mature death in the United States. . . . Cigarette smokers have a 70 percent greater rate of death from all causes than nonsmokers, and tobacco is associated with an estimated 320,000 premature deaths a year. Another 10 million Americans currently suffer from debilitating chronic diseases caused by smoking. In recent years, both the scientific and policy aspects of the smoking problem have become more compli- cated due to greater awareness of the interaction of smoking with other contaminants, such as asbestos in the workplace and community air pollutants.2 Increasing and Emerging Risks The declining incidence of age-specific mortality fails to reflect some especially pernicious risks that may be increasing or emerging, including the risks of crime, environmental contamination, and nuclear war. Since many emerging risks could have been selected for discus- sion by the committee, we have highlighted those that polling data suggest are of most concern to Americans (see Louis Harris and Associates, 1980~. cat- - r-~ ZThat awareness. as one committee member comments, does not always extend to an awareness of the relative impacts of smoking alone versus smoking in combination with other factors. Thus, if no one smoked tobacco, the proximate cause of 320,000 deaths a year, including most of the deaths of asbestos workers, would be eliminated. If, however, asbestos exposure had never occurred, only about 4,000 deaths would be eliminated, of which only a few hundred could be additional to those eliminated by stopping smoking.

OCR for page 1
7 Crime For many Americans, especially those living in inner cities, crime represents one of the most significant risks of daily life. Since the early 1960s, violent crime rates (murder, rape, robbery, and assault) have increased sharply, and survey data reveal that crime Is one of the risks of most concern e. O Americans. High and rising rates of crime impose psychic costs on individ- uals, which in turn impair the sociological health of communities. Moreover, (except for murder) these in- creasing crime rates are not directly reflected in longevity statistics. For many Americans, the risk of crime is especially serious and immediate. Environmental Contamination Although considerable progress has been made in the past 20 years in improving certain aspects of air and water quality, serious environmental problems confront us, and some environmental threats are increasing. Polling data indicate that Americans continue to be concerned with environmental degradation. That concern is seconded by other indications of continuing and new problems. For example: . . o While national emissions and ambient levels of sulfur dioxide and total particulates in the air have declined since the 1960s, there may even be increases in some geographical areas for sul- fates and fine particulates, two pollutants that may pose greater risks to human health than sul- fur dioxide and total particulates; U.S. emissions of sulfur and nitrogen oxides are combining with water vapor in the atmosphere to produce acid rain. As a result, the pH value of precipitation in the United States and other countries is frequently dropping well into the acid range, posing threats to fish, wildlife, and other organisms; The production of chemicals increased by a factor of 10 between 1945 and 1975, an indica- tion of economic growth and prosperity; this trend has also contributed to the growing prob- lem of safe disposal of hazardous wastes; i, ~e ~

OCR for page 1
8 The burning of fossil fuels by the United States and other countries is raising the amount of carbon dioxide in the atmosphere, which is now expected to significantly warm the global sur- face and thereby to affect regional climate patterns. These and other persistent, growing, or emerging environmental risks are not likely to be readily dis- cernible in historical or current mortality figures, due to long latency periods before adverse effects occur and the difficulties of distinguishing environmentally caused deaths from those caused by other factors. The increased manufacture and use of toxic chemicals and the rise in the incidence and the death rate from some can- cers have been a source of controversy. Whether there is a causal relationship between these two trends re- mains an issue involving large scientific uncertainties and considerable dispute among experts.3 Recent survey data indicate that "chemicals" are a major source of increased perceived risk among the pub- lic. m is finding underscores the importance of making the level of this risk more precise and of explaining it clearly. Nuclear War Although the vast nuclear stockpiles in the United States and the Soviet Union may have deterred a major military confrontation between these nations, such weapons have also created a fear of quick and immense destruction. The perceived risk of a nuclear calamity varies and seems to correlate with the existing state of tension between the superpowers. For example, after the signing of the Limited Test Ban Treaty in 1963 and of SALT I in 1972 the perceived risk was much lower than during the period of high tension created by the Cuban missile crisis. At this time, the uneasy relationship between the United States and Soviet Union combines with 3A committee member observes that although most ex- perts believe that there is no important causal rela- tionship, the issue is so important that they tend to be very cautious in their statements. The very '~edging" raises concern.

OCR for page 1
9 the volatility of the Mideast to make the possibility of a military confrontation escalating to a nuclear war a substantial threat.4 Although it is not directly measurable, the risk of nuclear devastation has probably increased since 1972 as a consequence of increased U.S./Soviet hostility and continued emphasis on nuclear weapons development in both countries. The emergence of China as a nuclear nation and the potential prolifera- tion of nuclear weapons in the Third World have added new dimensions to the threat of nuclear war. The risk of nuclear war is identified in this report for two reasons. First, it is crucial to identify those major risks that are not reflected in mortality in- dices. For example, the historical progress in increas- ing life expectancy does not take into account the per- sistent danger of nuclear war. Even if longevity is increasing, the probability of death may be constant or increasing if the probabilities of nuclear war are in- creasing. Second, the polling data on public concern about risks indicate the importance of global political instability, a concern that may take on greater signifi- cance due to nuclear armaments. The risk of nuclear war is fundamentally different from other daily risks of life for at least three rea- sons. Nuclear war threatens not only individuals but also current civilization and ecosystems. The risks of nuclear war are largely beyond individual control and may even be beyond one nation's control. Finally, rela- tively little is known about the likelihood of a nuclear confrontation's becoming an all-out exchange, the possi- bility of limiting escalation, or the degree of devasta- tion that would result from any nuclear war. These un- certainties provide an opportunity for the likelihood or consequences to be sensationalized or suppressed. ~e ~ Nuclear Power Polling data indicate that the risks of nuclear power are of considerably less concern to Americans than those 4A committee member comments that the increased mili- tary aggressiveness of the United States may reduce the probability of nuclear confrontation, because it signals our determination to remain firm.

OCR for page 1
10 of nuclear war (Louis Harris and Associates, 1980~.5 Moreover, these concerns differ not only in magnitude but also in kind. Rather than annihilation, the risks perceived by most Americans are more local, as in a meltdown catastrophe, or more subtle, as in genetic damage to future generations resulting from long-tenm exposure to ionizing radiation from nuclear plants or from repositories for their radioactive wastes. Also, large-scale nuclear power plants symbolize for some the imposition of a relatively new technology, which fosters an increasingly centralized and complex society in which the power of the individual is dimin- ished. Such concerns mean that debates on the course of nuclear power are inevitably flavored not only by tech- nical, economic, and political concerns but also, more strongly than in other issues, by philosophical ones. That debate continues to be vigorous in the United States and abroad. At one extreme, advocates argue that nuclear power provides an almost unlimited, low-cost energy resource essential in the transition from non- renewable fossil fuels to renewable forms, such as solar power. At the other extreme, opponents see nuclear energy as ushering in the potential for worldwide calamity threatening human survival; they link nuclear power with nuclear weapons and nuclear war, through the proliferation of new nuclear powers and through national and subnational terror; more important, they link the acceptance of nuclear power with acceptance of nuclear weapons and their use in another war. Those advocating nuclear power argue that the knowledge of nuclear fis- sion, even how to make bombs, is widespread. In this sense the nuclear genie is out of the bottle. If a country wants nuclear weapons, the proponents of nuclear power argue, it can acquire the fissionable materials needed via research reactors rather than power reac- tors. Also, despite the Three Mile Island accident, they further argue that nuclear power has proven safer and more environmentally benign than any other form of energy, predicting that this record can be maintained with continuing vigilance. Opponents challenge those . bA reviewer argues that perhaps 'the risks seen by the public are of the same kind, not different, and that this has been, in fact, one of the major stumbling blocks to the public acceptance of nuclear power."

OCR for page 1
11 predictions, arguing that with wider use the chance for accident and for radioactive contamination will inevi- tably rise. Lay people as well as scientists are split on those issues. Even though public referenda have shown that the bulk of public opinion lies well within the extreme points of view given above, public apprehension about nuclear power remains high. These uncomfortable strains in the development of nuclear power undoubtedly contrib- ute to apprehension in a significant sector of society and hence to its belief that life is becoming riskier, despite mortality statistics to the contrary. _ RISK PERCEPTION Anxiety about risk is a product of people's beliefs and attitudes. People often form such beliefs and attitudes on the basis of incomplete and often biased information using fallible modes of inference, which sometimes result in systematic distortions and misperceptions of reality. These heuristics, although sometimes useful and convenient, result in systematic distortions and misperceptions of reality. For example, scientific progress identifies new risks, which are reported to the public. There are reports on the risks of the Three Mile Island accident hazardous chemical wastes in the Love Canal, DC-10s, benzene, saccharin, asbestos, tampons, PCBs, the pill, . recombinant DNA, nuclear waste, and so forth. These reports may make certain risks more memorable or imagin- able, thereby increasing perceived probabilities of risk regardless of the total scientific evidence about or trends in actual indices of risk. If this phenomenon is occurring on a large scale, it may explain part of the rising public concern about risk. As people become better informed about various health and safety risks, levels of concern can be expected to rise.6 This process is accelerated by scientific advances that detect more and more previously unknown sources of risk. tA reviewer remarks that better information can sometimes indicate decreased risks. While perceptions of existing, real risks change, scientific advances may introduce new risks and alleviate old ones.

OCR for page 1
36 reliability and acceptance of uncertainty. For example, in a recent risk assessment--the one deal- ing with ozone depletion--epidemiology, dose- response extrapolation, chemical modeling and atmo- spheric modeling, were employed in sequence in order to estimate the potential increase in ultraviolet light-induced skin cancer and malignant melanomas possibly resulting from continued release of chloro- fluorocarbons. tne problems involved in compounding the uncertainties by combining theoretical simula- tion models, laboratory reactions, clinical observa- tions, and epidemiological correlations have not really been examined by the risk assessment litera- . . and are an important research task. Lure . Value Trade-Offs and Ethical Considerations in Risk Evaluation Risk reduction is not society's only goal. Many of the risks in society are worthwhile or acceptable, at least in the sense that trying to reduce them would, all things considered, make life worse. In some cases, society may choose to increase certain risks in order to attain other important goals. Certainly individuals do so. Hang gliding is risky and could be considered un- necessary, yet it is popular. There also may be compet- ing risks. Lowering one may raise another, or lowering one risk may be so costly to industry, and eventually to consumers, that other, more effective actions cannot be afforded. However, even if an action or a decision not to take an action raises the net probabilities of ad- verse consequences, it may have other, compelling bene- fits. On balance, society may choose an option that does not minimize long-term net risk. The broad range of objectives that should be weighed in deciding how to cope with a risk can be suggested by a specific example: an analyst trying to evaluate the consequences of some regulation intended to protect human health or safety. The evaluator may want to ponder various kinds of health effects, including how many people would be affected, not only in the entire population but also in sensitive groups. How much they would be affected (in terms of mortality, morbidity, pain, suffering, discomfort, and perhaps anxiety), who they are (by age, income, occupation, geographical location, etc.), and when they would be affected may

OCR for page 1
37 also be relevant. In addition, the evaluator may be concerned about the degree to which the risk was volun- tary or involuntary and about whether the effects would be scattered across the country or concentrated in some small group of people. But there are other measures to consider. The evaluator should be interested in effects on nature, economic growth, productivity, innovation, business competition, the distribution of income, public satisfaction with government, and the quality of busi- ness and personal decision making. Pconomic costs--and for whom--certainly matter, as might enforcement costs and political costs. These and other considerations such as aesthetics, due process, and international rami- fications may all be highly uncertain, subject to reas- sessment as new information is gathered, the target of wide disagreement, and stretched out over time; the evaluator might want information about this. Further- more, an evaluator might want to know how identifiable the victims are, the type of injuries to be expected, the number of people affected, and how accountable the decision maker will be as well as what the possibilities are for delay, experimentation, flexibility, and adapta- bility. There is more--this is just a partial checklist that could be greatly expanded. This list is not appropriate for all kinds of risks; quite a different list would apply, for example, to arms control. Furthermore, many decisions about risks will not contain many of the items in the checklist above. It is clear, however, that such lists will frequently be long and that some very difficult trade-offs among com- peting objectives will have to be made. Much of the disagreement today about how to cope with a variety of risks stems from uncertainty about their magnitude: how hazardous is benzene or saccharin or ionizing radiation? Extrapolations from observable to nonobservable conditions, either in dose level or from animals to humans, will probably remain conten- tious. If, however, contrary to all expectations all such scientific imprecision could be resolved, contro- versy would still abound and perhaps would be even more vehement. Instead of a heated argument about linear versus nonlinear extrapolations from high to low dose rates, the dispute might be about the role of government in a free society. A few simplified examples of some of the most per- plexing problems involving ethics and value trade-offs are sketched below.

OCR for page 1
38 Bow much of society's limited resources should be allocated to life-saving activities versus other pres- sing social concerns? If you had to make the uncomfortable choice of saving an anonymous 1-year-old's life or that of an anonymous 20-year-old or 60-year old, which would you choose? How important is the psychological well-being asso- ciated with clean air and blue skies compared with vari- ous levels of economic well-being? How should on an asthma attack suffered by a 30- year-old be compared with a bout of emphysema suffered by a 60-year-o1d? How should our society decide whether to save the lives of inhabitants of North America 100 years from now versus saving starving Sahelians now? Do we have the responsibility of maintaining ecolog- ical balances for nature's sake rather than--or in addi- tion to--for people's sake? To what extent should society be willing to impose . . . ~ ~ _ costs and risks on a few members of society in order to benefit most members of society?l5 When should heroic measures be taken to prolong life? When should governments restrict individuals from self-hazardous behaviors? These long-standing issues of value are common in risk and decision making. While science and profes- sional analysis may contribute to a more informed debate concerning these trade-offs, they cannot resolve them; they may, however, help to define and structure those value trade-offs. This perspective is important because it highlights the limited albeit important role of science and professional analysis in resolving disputes over decisions about acceptable risk. 5A variation of this is: To what extent should society be willing to impose small costs and risks to a large number of members of society in order to greatly benefit a few members of society? There would be a large societal consensus on the answer if the above question were changed to say, imposing great costs to many that modestly benefit a few.

OCR for page 1
39 THE ROLE OF SCIENTISTS AND SCIENTIFIC INSTITUTIONS Many scientists are active in public debates about risk and decision making. As citizens they should be in- volved in the political process. Indeed, it can be argued that scientists who can understand the technical intricacies of complex problems have a special obliga- tion to speak out on controversial issues. But there is a price society pays for this openness. Passionate claims by equally eminent scientists on diametrically opposed sides of a policy debate may bewilder lay people. me media may exacerbate this problem by focusing attention on extreme opinions that are accom- panied by ringing rhetoric. It is hard, for the media and for the general public, to sort scientific judgments from what is primarily value-based or political opin- ion. In some cases scientists cloak controversial value judgments in scientific jargon, both consciously and unconsciously. As a result, the prestige of science and public confidence in science suffers. This dilemma has profound implications for science. It may affect--some say that it already has affected--funding for science, the selection of professions by college students, and the willingness of scientists to participate in the policy process. Furthermore, scientific progress may hold one of the keys to better and healthier lives for all, especially those in the disadvantaged, developing world; ironically, scientists, although meaning well, by engaging in heated debates involving unscientific value ~ . . . _ - Judgments may be undermining the tuture potential ot science for amelioration of the world's ills or for coping with the critical problems created by overpopula- tion, the depletion of resources, and the degradation of the environment. (Some committee members feel that the above remarks understate the gravity of the problem; others find them a bit melodramatic but worthy of re- flection.) Scientists engaged in debates about risk and deci- sion making should, whenever possible, attempt to say where their scientific expertise ends and where their personal value judgments begin. Of course, a scientist may not know where the boundary between the two falls; scientific expertise and value judgments may fuse, but an attempt should be made to disentangle the two. Of course, there may be a continuum of states of fuzziness, but an attempt should be made to communicate this fuzzi- ness. Further, scientists in public debate should, . -- .... ~ I. -

OCR for page 1
40 whenever possible, refer to credible evidence not sup- porting their views. Committees of specialists working under the aegis of prestigious scientific institutions who have been asked to perform some risk analysis have a responsibility both to the committee's client, the public, and the commit- tee's parent institution. These responsibilities are jeopardized when tasks are accepted that are inappropri- ate for the committee and its parent institution. For example, a committee of physical scientists sponsored by an engineering institute may not be the appropriate com- mittee to recommend policies for balancing concerns about health and safety risks with value trade-offs and political constraints. Some scientists on the committee may be knowledgeable about policy concerns, but if their expertise is sought and their institution's imprimatur is given on the basis of their professional expertise as physical scientists and not as policy analysts, then it may be inappropriate for them to make pronouncements that extend beyond their disciplinary expertise. Cer- tainly if they feel compelled to offer policy recommen- dations, such excursions should be clearly labeled in their report. Outside the committee, the members have the right and the duty to speak out on the issues, but the committee is not serving its parent institution, the . . . . . . Drosoer scenic community, or the public if it ex- pands its consultative role from the descriptive assess- ments of scientific phenomena to the prescriptive evalu- ations of policy choices--if that was not their assigned task.16 Committees are often specifically asked by their clients to suggest recommendations for policy choice "all things considered." Depending on the composition of the committee this may not be unreasonable, but com- mittees should be careful lest they become used and manipulated in a political process. Sometimes a good decision may result from shifting the burden of choice from a highly politically charged regulatory commission to some quasi-scientific body of experts, but the shift 16A commentator remarks: The scientific community needs to figure out ways to defend scientists who par- ticipate in these debates and then find themselves having their good faith questioned. This constitutes a significant disincentive for reputable scientists to become involved.

OCR for page 1
41 may be made at the expense of confusing the role of the parent scientific institution. Scientific institutions should and sometimes do inform a potential client that its interests (and society's broad interests) would be better served if the scope of the assignment is limited to nonprescriptive, nonevaluative scientific assessments. To talk about assessment and evaluation--their sepa- ration and integration--in risk analysis, we purposely draw sharp lines around each. ~ ~ ~^ ~ These lines are difficult to draw and members of many committees do not think deeply about these distinctions. Even if the agreed-on scope of a committee is to concentrate its attention on descriptive science, the committee may nevertheless slide into prescriptive, evaluative elements of an analysis, not as a conscious decision but as a natural consequence. Our suggestions are not stark. We are not saying, 'tDo not mix evaluation with assessment"; rather we are saying, "Be conscious of what you are doing. If evaluative components are prominent in your report, was such an emphasis intended? Has the committee been appropriately chosen for this task? Did the committee spend enough time considering the dynamics of choice, value trade-offs, and political and legal constraints before they arrived at their policy conclusions?t' These questions suggest there may be a conflict of interest between the needs of a client and the reputation of scientific institutions. A troublesome and not uncommon conflict is the degree to which political feasibility should be con- sidered in risk assessments. A committee may opt to consider an issue purely on technical and economic grounds only to have its judgments labeled as polit- ically naive; or it may indeed consider the political climate in weighing different options and be accused by its sponsor of overstepping its charge. A reasonable path between these two traps is for the assessors to clearly distinguish between judgments based on technical and economic grounds and those based on political con- siderations. That, again, is a matter of competent integration of assessment with evaluation. SOME ISSUES IN RISK ASSESSMENT There is one major source of controversy that plagues many risk assessments: How far should speculative numbers be pushed and combined? If benefits are

OCR for page 1
42 quantified, then there is the criticism that the 'Ward drives out the soft," that aesthetic, fragile values are ignored. Yet if these soft values are subjectively and numerically assessed, the resulting numbers are then subject to attack. If there is a vitrolic debate about whether some intangible benefit should be evaluated as worth $10,000 or $20,000, it may be easier to do the allegedly objective thing and leave it out altogether --i.e., to make it "zero." Critics complain about the failure of formal techniques to incorporate the intan- gible hunches and intuitions of experts; but if these hunches are incorporated by numerical scaling techniques (e.g., by subjective probabilities), then they can be easily ridiculed by skeptics. In the debate about how far to quantify, as in most long-standing debates, there are errors of two kinds in the balancing equation: A false sense of precision with numbers may give the impression that more is known than is really known; and a false sense of imprecision with- out numbers may give the impression that less is known than is really known. Risk assessments deal with uncertainties: some are based on copious amounts of relatively uncontroversial statistics (such as assessments of motor vehicle risks); others are based largely on reasoned speculation (such as assessments of the possibilities of sabotage of a nuclear power plant). In the former type of cases, nearly everyone would agree about the relevance of sta- tistics and estimates of probability. In the latter type of cases, although most people might have a hard time seeing how numbers can be used at all, a few people might want to numerically scale the judgments of ex- perts. hey would argue: We cannot ignore the problem of making assessments because decisions must be made, even if the decision is "no change." The problem is how to be appropriately precise about the state of impreci- sion. Some might paraphrase this as how to be appropri- ately imprecise about utter chaos. It is important for decision makers to know not only what is known but also what is not known.l7 70ne member of the committee points out that this is the procedure now recommended for dam failures, for which there is little theory and limited experience but a great deal of expert judgment.

OCR for page 1
43 There are numerous arguments for and against the use of numerical probabilities in risk assessments. But lurking in the background is a question: If you do not use probabilities, then what do you do and how will it respond to policy needs? Should experts, for example, report their subjective judgments about uncertainties when such judgments cannot be formally based on objec- tive data or on well-articulated theoretical models? All agree that if subjective judgments are reported, they should be clearly labeled as such and accompanied, insofar as possible, by supporting arguments. However, some would shun subjective statements in scientific documents altogether, especially the mixing of subjec- tive and objective judgments. It should be noted that our reference is not to subjectivity in value and poli- tical judgments, but to expert judgments about uncer- tainties, such as the uncertain severity of a toxic substance. In risk assessment, these philosophical debates about the very foundations of statistical inference have some profound, practical implications. In many investigations of the uncertainties of risks, the scientific facts speak for themselves without the need for expert synthesis; just marshalling and exhibiting them is sufficient advice to the policy maker. In other cases, however, the facts pull in different directions with different degrees of credi- bility, and if all this raw information were accurately and completely reported, it would overwhelm any policy maker. Policy makers may want experts to interpret and synthesize those facts, so that they can incorporate values and political concerns to arrive at a balanced decision. But synthesis cannot simply be computerized. It remains an art form, and judgments--expert judg- ments--have to be factored in. Yet ideally the policy maker wants the specialists to stick to their craft; the policy maker wants to learn what the specialists know about the scientific uncertainty and not about their value preferences or their political perceptions. The object is to communicate information from assessors to evaluators in a manner that will be useful in making policies and decisions. The synthesis of assessment information is profoundly influenced by whether these reports are cast in a subjectivist or an objectivist mode. Subjectivist reports may be more responsive to the evaluators' needs but may have a more difficult time in peer review.

OCR for page 1
44 Specialists disagree for many reasons, and some of them persist in disagreeing after they listen to each other at length. Perhaps there are communication bar- riers, or perhaps they talk in different paradigmatic languages, or perhaps they cannot articulate how they really feel, or perhaps they are distrustful of each other and are acting strategically. But disagree they often do. The question in what to do about such disagreements. A good case in point is a recent NRC report on the effects of radiation (National Research Council, 1981~. In extrapolating carcinogenic effects from high doses to low doses of radiation, the experts are divided on the appropriateness of three underlying extrapolatory models. In this case the experts can agree about what they disagree about--at least at the level of the choice of a model. The answer is not critical for all policy decisions--all models, for example, lead in some cases to the same policy con- clusion. But if not, what then? Can they agree on why they disagree about the appropriateness of the three models? Given that they disagree about fundamentals, can they agree on a compromise assessment report to give to the policy maker? How should groups report their ~ r - - -r disagreements? A proper separation and integration of assessment and evaluation can founder at this juncture. Indeed, it may be so difficult for an assessment group to agree on a synthesis that it may be necessary to fuse assessment and evaluation. Government agencies frequently ask scientific organ- izations to perform rink analyses by using the committee arrangement. If the task of the committee lies exclu- sively with risk assessment, then the membership of the committee should be structured primarily to yield a balanced portfolio of relevant scientific and methodo- logical skills rather than a balanced portfolio of policy viewpoints. This procedure would encourage open, honest, and nonadversarial interactions and discourage strategic, posturing regarding policy values. Some caveats apply to this, however. For one, the committee's membership and its capacity for calling on different sorts of experts should remain flexible as its perceived needs evolve during the study with deepening insights. In addition, as we emphasized earlier in com- menting on the difficulties of integrating assessment with evaluative components, there is a need for active, iterative interchange between the assessment committee

OCR for page 1
45 and those who will use its report in formulating poli- cies and decisions. Otherwise the committee may solve the wrong problem, grapple with issues irrelevant to the evaluator, or fail to gather the most important infonma- tion. Finally, it is not always easy in selecting com- mittee members to separate scientific competence from policy involvement; indeed, that competence may be due In part to Involvement In policy. Rigidity in selecting members purely for scientific competence may result then in a committee tilted toward one side of a policy de- bate. In that situation, corrective compromises may have to be made or the tasks of the committee modified. If these compromises are not made, then the public might, incorrectly, perceive ~ ~~ biased. the committee s report as Assessment reports commissioned by clients for the policy process are generally made public and reviewed by the scientific community. Peer reviews, whether for- mally commissioned or individually motivated, serve many functions: They may help potential users judge the calibre and integrity of the report; they partially motivate the authors of the report to be more respon- sible; they help to weed out errors; they may stimulate others to accumulate new evidence that would be relevant to any reassessment; and they focus attention on the weak links of the analysis. The peer review process is most effective when there is a clear demarcation of analytical tasks. When risk assessment and risk evaluation activities are mixed, it is especially difficult to orchestrate an organized review by appropriate specialists. Yet when risk as- sessment is conducted separately from evaluation, an elaborate and flexible review process can improve the quality of assessments and evaluations. For example, not only can there be reviews of the primary assessment report, but also, in a somewhat adversarial style, there can be reviews of the reviews. Funding agencies for risk assessments can incorporate a dynamic review pro- cess into funding decisions. It might be advisable to reserve substantial monies for peer review or possibly to fund two independent risk assessments, depending on the importance of the problem at hand. A final note. It is very hard to separate assess- ment from evaluation. And it is hard to do good risk assessments; there are not many examples of such assess- ments that blend well into the evaluation tasks. How- ever, the technique of deliberate separation is fairly

OCR for page 1
46 new. When done well, the results often repay the diffi- culties. Egregious errors are filtered out; high- quality reviews are more 1 ikely; and the overall stan- dards o f risk analyses are raised .