Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 Recommendations for Improving Risk Communication Drawing lessons from the available understanding about the na- ture and problems of risk communication, we present four sets of recommendations in this chapter: (1) recommendations that pertain to the processes that source organizations use to generate decisions, knowledge, and risk messages; (2) recommendations that pertain to the content of individual risk messages; (3) a call for a "consumer's guide" that will enhance the ability of other groups or individuals to understand and participate in risk management activities; and (4) a brief summary of particular areas for which additional knowledge is needed to resolve current problems of risk communication. We have attempted a focused search. The committee faced a central dilemma about how detailed we could expect to be in meet- ing our charge to discern practical lessons for practitioners. Given the breadth and diversity of the general topic of risk communication, any attempt to Took for lessons that apply to all forms of risk com- munication would constrain us to a discussion so general that any particular reader would gain little insight. On the other hand, a de- tailed "cookbook" for particular situations would fail to advance the broad national discussion that is now needed. We have accordingly sought a middle ground, electing to narrow our scope in two ways. First, we have elected to focus on certain forms of risk commu- nication. The term "risk communication" can cover a vast range of 143
144 IMPROVING RISK COMMUNICATION actions, from casual telephone calls between two experts to book- length reports meant for the general public. Our main subject in this chapter is formal risk messages intended for audiences that in- clude nonexperts. Included, for example, are press releases, material prepared for an open meeting in a community or a formal meeting with representatives of interested outside groups (e.g., a local public meeting about siting a facility), a government agency's public ex- planation of a decision it has made, a brochure for citizens concern- some aspect of public health (e.g., an AIDS pamphlet), package inserts for prescription drugs, and risk summaries prepared by ex- perts within an organization for the use of their (less-expert) superi- ors. We recognize that some of our recommendations may have less relevance for other very important, but less formal, varieties of risk communication. Second, we have directed our recommendations to just two of the many types of risk-managing organizations that are discussed in other parts of our report: namely, government agencies and large private corporations. Again, this choice of emphasis is not intended to imply that other communicating organizations and individuals- small firms, citizen/consumer advocacy groups, and so on-are not important. In fact, many of the points we raise doubtless apply to them. We chose this narrower range of organizations because they are most directly involved in many of the best known and most controversial cases, the committee members have greater knowledge of their experiences, and we are convinced that improvements by these organizations would both contribute substantially to easing the national problem and provide models for other organizations. Our objective, then, is to improve risk communication, particu- larly as practiced by government and large corporations. What do we mean by "improve"? We mean that solutions sometimes ad- mittedly only partial solutions are put in place for the range of problems identified in the previous chapter. We emphasize in par- ticular that we have tried to fashion recommendations that, while addressed to government and large corporations, will attack the prob- lems of recipients as well. Our goal is not then to make those who disseminate formal risk messages simply more effective by improv- ing their credibility, understandability, and so on such an approach might serve their interests but could well degrade the overall quality of risk communication if it meant that they would merely advance their viewpoints with more influence. "Improvement" can only oc- cur if recipients are also enabled to solve their problems at the _, ~
RECOMMENDATIONS 145 same time. Generally, this means obtaining relevant information for better-informed decisions. We have also focused our recommendations on measures that will help those groups meet the criteria we have set out above for suc- cessfuT risk communication. In reality, of course, many organizations have other criteria for success, such as whether messages convince recipients to act In a manner that the risk communicator desires. We have not chosen to recommend actions to hale or~z~ni~z~t.ir~ne mart those other goals. In recommending steps to be taken by government entities, we have necessarily focused on the respective roles of citizens, private groups, and government in a democratic society. Controversies about risk communication often turn out to be basic debates about the lim- its of governmental accountability, legitimacy, and authority. The ~ _ _ 1 _ ~_ ~. . . ~ < -- - <3-F^~- eLA40-U goal or our recommendations is not to alter American democratic institutions but to make them work more effectively. Two points need to be emphasized about accountability. First, our society has elaborate and politically responsive procedures for assigning respon- sibilities for making government risk management decisions. Once a government agency has received that responsibility, it must retain it. This places inherent limits on what agencies can do in discussing risk issues with citizens, because they cannot share responsibility with outside groups; they must remain publicly accountable. Second, ac- countability increasingly implies an affirmative duty to interact with interested and affected outside parties in reaching and explaining individual policy decisions. Although citizens" and the groups that undertake to represent their interests are not required to Dartic _ · ~ ~ ~· . · mate in such interactions, solving problems of risk communication becomes much easier if they do, and government needs to ensure that the opportunity to participate becomes routine. Implementation of many of our recommendations rim or ganizational resources of several kinds. We are aware that such resources will not be adequate in many instances. One r~.c:~,rr~ in particular time-is crucially lacking for some of the most difficult risk communication efforts, as when emergency conditions leave no possibility of consulting with outside organizations or assembling complete factual information. Other recommendations require staff resources and the capacity to conduct specialized analyses, both of which may be in short supply in some organizations. When resources are so constrained, our recommendations may well best serve as a reminder of the full set of factors that should be accommodated,
146 IMPROVING RISK COMMUNICATION although the form of accommodation may fall short of what we recommend. Our recommendations are based on our understanding of the growing literature of studies of risk communication and risk mes- sages and on committee members' diverse experience with specific instances of risk communication. Before we list our recommendations, we would like to draw at- tention to three general conclusions that we have made: Conclusion 1. Even great improvement in risk communication wit! not resolve risk management problems and end controversy (al- though poor risk communication can create them}. Because risk com- munication is so tightly linked to the management of risks, solutions to the problems of risk communication often entail changes in risk management and; risk analysis. There is, unfortunately, no ready shortcut to improving the nation's risk communication efforts. The needed improvement in performance can only come incrementally and only from assiduous attention to many details. While it is important to improve risk communication practices, no one should expect such improvements to end public controversy over risk management. Risk managers should understand and accept that, even when they have done all they can to ensure the integrity of their risk messages, public skepticism of their motives and their honesty will likely persist. They should appreciate that, particularly in recent years, distrust has been institutionalized in our country. While it is important for most risk managers-especially those in the government to avoid distortions in their messages, they should expect that many audiences will continue to assume that bias is present. We have discovered no sweeping broad-spectrum remedies for the problems of risk communication described in Chapter 6. Many will be solved only over the long term and only by sustained ef- fort. Many of the institutional problems we identified in the previous chapter fragmentation of authority, legal constraints, and so on- reflect social decisions about how risk management should be con- ducted. Such decisions are inherently, and appropriately, political in nature. Risk communication might well be improved if certain con- textual constraints were changed or removed. However, such reforms would also create other advantages and disadvantages that are well beyond our capacity to evaluate in this study. Thus we are left with a more modest, ant] necessarily incremental, set of available remedies.
RECOMMENDATIONS 147 The source organization's problem of achieving credibility pro- vides a good example. An organization's credibility can be quickly lost, as illustrated in the case of the EPA in the early 1980s, when many observers came to believe that one of EPA's leaders' highest goals was to dismantle regulatory programs. In contrast, credibility is gained (or regained) only through a sustained effort to be re- sponsive to audience concerns and to be accurate, open, and honest in disclosing essential information. Thus we are led to recommend concurrent attention to several factors in managing the risk commu- nication process and in formulating particular risk messages. No one of these measures, alone, is enough. An underlying reason for this is that the problems of risk com- munication are rooted in risk management practices and procedures. Because of this, several of the measures we recommend call for adjust- ments in the source organization's procedures for risk management and for analyzing risk issues. For example, we call for more interac- tion with audiences and intermediaries while the source organization considers risk management alternatives, and we suggest how formal risk assessments should be scoped, reviewed, and presented. We have explicitly addressed many of our recommendations to risk managers precisely because they are the individuals within an organization who can provide the needed coordination of risk communication, risk management, and the assessment of risk and risk control. O ~ Conclusion 2. Solving the problems of risk communication is as much about improving procedures as improving content. Risk man- agers need to consider risk communication as an important and inte- gral aspect of risk management. In some instances, risk communica- tion will, in fact, change the risk management process itself. . It would be a mistake to believe that better risk communication is mainly a matter of crafting better messages. To enhance credibil- ity, to ensure accuracy, to understand recipients and their concerns, and to gain the necessary insight into how messages are actually apprehended, one must ultimately seek procedural solutions. Thus we devote much of this chapter to matters of process. There may be many cases in which problems of credibility, potential contro- versy over value judgments, and diverse audiences reduce the risk communication task to a simpler matter of making messages clearer, in themselves. We do not believe that the national frustration over risk communication practices derives from failures in such "simpler" cases and therefore have not addressed simpler cases in any detail.
148 IMPROVING RISK COMMUNICATION Risk managers cannot afford to treat risk communication as an afterthought. One of the root problems in risk communication is that, perhaps due to organizational imperatives and tradition, risk management has too often been treated as a sequential process: (1) the organization's technical experts assess a risk and explore options, (2) a risk management decision is made, (3) a message is internally prepared, and (4) the message is sent to outsiders. Risk communication is thus regarded as a subsidiary activity. The importance of risk communication has only recently become apparent, and even the most progressive risk managers are only now beginning to adjust to the realization. Improvement of risk communication requires that the organizations that disseminate risk messages become simply more deliberate in their communication efforts. At their best, risk communication efforts can be expected to af- fect the risk management process itself. Considerations of risk com- munication might, for example, determine what kinds of analyses of risks and benefits are performed, how risk assessments are summa- rized, what options are explored, and what people are consulted in exploring possible courses of action. Risk communication requires its own specialized expertise and deliberate planning and evaluation. Senior managers need to devote attention and time to managing risk communication efforts per se. It is a mistake to simply consider risk communication to be an add-on activity for either scientific or public affairs staffs; both elements should be involved. There are clear dangers if risk messages are formulated ad hoc by public relations personnel in isolation from available technical expertise; neither can they be prepared by risk analysts as a casual extension of their analytic duties. Conclusion 3. Two broad themes are apparent in the extended list of recommendations: that communication efforts should be more systematically oriented to specified audiences and that openness is the surest policy. Both the management of the process of formulating risk messages and the content of risk messages should be systematically oriented to the intended audience. The most effective risk messages are those that quite self-consciously address the audience's perspectives and concerns. Similarly, the best procedures for formulating risk mes- sages have been those that involved interactions with recipients and that elicited recipients' perceptions and needs.
RECOMMENDATIONS 149 A central premise of democratic government the existence of an informed electorate-implies a free flow of information. Suppres- sion of relevant information is not only wrong but is usually, over the Ton ger term, also ineffective. Risk messages should be explicit about current knowledge of the subject risk but also about the lim- its of that knowledge and the existence of disagreement among the experts or others. The long-term improvement of credibility, in par- ticular, depends on openness. Several of our procedural and content recommendations are intended to foster openness and to promote openmindedness about outside viewpoints. MANAGEMENT OF THE PROCESS Much recent concern about risk communication has centered on questions of message content. Failures have frequently been at- tributed to the inability of the audience to comprehend complex technical issues and to the tendency of risk messages to be badly written. This view would lead one to seek solutions in the design of better risk messages themselves. Our assessment has led us to believe that longer-term solutions are equally likely to involve attention to and changes in the process by which risk management decisions are made and explained. There are two basic reasons for our emphasis on process. First, when lessons about message content are identified, the operational question becomes one of ensuring that those lessons are systemati- cally followed. Procedural safeguards provide the best assurance of routine compliance. Second, and more important, it is increasingly clear that content and process are not easily separated, particularly on the crucial matter of appearing credible. If recipients believe the process is flawed for example, if the communicating organization is known to ignore or reject certain facts, viewpoints, or options they are likely to doubt the message, even if it is, in fact, technically competent. This section is addressed to risk managers- those senior officials who have the overall responsibility of determining their organiza- tion's action. These risk managers also oversee the preparation of risk assessments and risk messages associated with the action to be taken. We identify four process objectives that are key elements in improving risk communication: goal setting, openness, balance, and competence. We note that these objectives are general in nature.
150 IMPROVING RISK COMMUNICATION Different management styles may work best for different managers in particular situations, in pursuit of these four objectives. Setting Realistic Goals Some past deficiencies in risk communication efforts have arisen because risk managers have not appreciated that risk communication needs to receive deliberate management attention. Until now, risk communication efforts have all too often been pursued with implicit or impractical objectives within the source organization. Risk comImmication activities ought to be matters of conscious design. Practical goals should be established that explicitly accom- modate the political/legal mandates and constraints bounding the process and the roles of the potential recipients of the organization's risk messages. Explicit consideration of such factors encourages real- istic expectations, clarification of motives and objectives (both within the source organization and among outside groups and individuals), and evaluation of performance. Consideration of these issues of practical goals and impediments to their achievement may be the only way for managers to reach re-~ alistic expectations. Otherwise, source organizations may set them- seIves up for frustration and, if naive or insensitive programs result, for disrespect among recipients that can only aggravate any preex- isting tensions about how the risk should be managed. Effective program management is enhanced by setting explicit objectives. This is especially important with respect to risk commu- nication because of the difficulty of assessing the effect of messages. A cornerstone of systematic risk communication goals is a realistic review of the political and legal context of the communication ef- fort and the risk management decisions to which it relates. What is one empowered to do? Can messages properly attempt to induce recipients to take certain actions or can they only transmit neu- tral information? Who must receive the information? What level of understanding (if any) must be assured? How active a part can interested and affected parties be allowed to play in the risk manage- ment process? Analysis contributing to goal setting provides a way to articulate the basic premises for action and a basis for evaluation of performance. Such analysis sets the general context for a risk communication effort. It needs then to be translated into operational objectives. For example, how many people should receive the message? What
RECOMMENDATIONS 151 changes (if any) should be observed in recipients' beliefs or actions regarding the risk? Will recipients be motivated to listen? Will they rely on other, possibly contradictory, sources? Realistic assessment of factors affecting message preparation, transmission, and receipt can be an important contribution to an organization's effective par- ticipation in the risk communication process. Safeguarding Openness In many cases risk communication efforts have founclered because public trust and credibility were damaged because risk management was conducted behind closed doors or because of a patronizing atti- tude toward interested outside groups. Risk communication should be a two-way street. Organizations that communicate risks should ensure elective dialogue with poten- tially affected outsiders. This two-way process should exhibit: . a spirit of open exchange In a common undertaking, not a series of 'canned briefings discussion should not be restricted to technical 'nonemotional issues-and early and sustained interchange, including the media and other message intermediaries. Openness does not ordinarily, however, imply empowerment to de- tee ne the host organization's risk management decisions. To avoid misunderstanding the limits of participation should be made clear Tom the outset. O:~1 ~L _ 1 ~- 1 ~1 . . . · . . .. . l~l;Sx 1lla~lag~r~ snoula resist one temptation to close their pros cesses to outside scrutiny and participation unless, as is rarely the case, extreme conditions warrant secrecy. As a practical matter, problems of risk communication for many past cases seem most pro- nounced when risk communicators have not appeared to value open- ness. In addition, many of the cases that were resolved relatively effectively were marked by openness. Openness thus has practical benefits both for the organization that manages risk and for outside participants, but there are deeper reasons for it. Openness is highly valued in a democratic society like ours because public accountability is a central element of our political culture. This is particularly true for organizations that are respon- sible to an electorate or that are charged with a public purpose, but private organizations are hardly immune in contemporary America. The fact that ours is a democratic culture means that there are strong negative sanctions in public opinion for evidence of secrecy. When
152 IMPROVING RISK COMMUNICATION governments or corporations can be found guilty of withholding in- formation, they commonly find themselves severely condemned, and their credibility is damaged for some time, regardless of the content of their risk message. Thus openness may be seen both as a matter of principle and a matter of practical wisdom for operating in a culture where many others take openness to be a matter of principle. Openness may take diverse forms in diverse risk management settings. When a government agency considers issuing a regulation, it can involve representatives of interested and affected groups in discussions of the rationale for action, quantitative and qualitative indications of the subject risk, available alternatives, and other fac- tors affecting its choice. If an organization undertakes to advise the general public of a risk associated with personal behavior (e.g., diet, sex), it can involve representatives of the intended audiences in discussions of the need for risk messages and the best ways to compose them. If a corporation decides to locate a new facility in a community, it can draw community groups into discussions of the nature of risks presented by the facility and take steps to control such risks. Risk messages will prove much more Biscuit to convey when recipients believe they were excluded from risk management decisions that affect them. Openness also provides an opportunity for risk managers to re- ceive important information from outside the organization relevant to their risk management decisions, as is amplified in the later dis- cussion of competence. Effective Dialogue The most productive interactions are those that treat outside parties as fully legitimate participants, so that two-way exchange occurs. If the host organization conveys the impression that it is meeting with groups simply to diffuse outside concerns, or to edify "uninformed" lay risk perceptions, this goal cannot be met. If mutual trust is established, the host organization will benefit from fresh ideas, will understand better how its formal risk messages will be perceived, and will be able to incorporate needed adjustments to messages earlier than if opposition forms in response to a message. Participating organizations will have a chance to understand the basis for action and to determine for themselves the degree to which the risk decision and the associated risk message are based on full and open-minded consideration of available knowledge and the full range of alternative actions.
RECOMMENDA TIONS 153 Eliciting participation is not simply the passive provision of access to the process of forming risk messages. Many outside groups have had frustrating experiences in which their views have been elicited but not listened to. An example is the holding of pro forma public hearings, which frustrated participants later fee} should have been labeled as "talkings," not hearings, from the host organization's apparent lack of attention to points raised. Active effort should be applied to identifying the full set of interested and affected groups and ensuring that the full range of potentially contending viewpoints is apprehended. The risk manager should ensure that those in the organization have come to understand: . what the participants know, believe, and do not believe about the subject risk and ways to control it; . what quantitative and qualitative information participants need to know to make critical decisions; and . how they think about and conceptualize the risk. To accomplish this, those within the organization who interact directly with outside participants should be good listeners. They should not make facile prejudgments about what people think and know and which options they will prefer. They should be prepared for skepticism, antagonism, and hostility. They should respect the legitimacy of subjective, as distinct from coldly analytic' responses. They should not be surprised if people are more interested in matters of trust, credibility, and fairness than in the technical details of risk estimates and risk reduction options. They should not expect outside participants to know, or to necessarily accept, the legal or other practical boundaries that constrain the risk decision. Risk managers should expect, and not resent (or appear to re- sent), skepticism about their motives in establishing more open pro- cedures. They should understand that the fear of co-optation may impede trust, at least initially. The job of interacting with outside participants should not be delegated to lower-level staff. Those with the power to make the decisions under discussion need to be directly involved in face-to-face dialogue, at least for the major issues, for this provides convincing ev- idence of the organization's sensitivity to the viewpoints of interested and affected groups. In some cases it may be advisable to formalize the participa- tion, for example, by forming a citizen advisory group. Such a move would signal an organizational commitment to continue to listen en cl
154 IMPROVING RISK COMMUNICATION to heed. Representative sample surveys also can help identify what people know and how they feel, what they think their choices are, and their responses to new information. Such surveys could constitute a valuable contribution to the openness of the overall process. Early and Sustained Interaction The best form of interaction is that which begins at an early stage and continues from then on. If outside groups are brought in very late, they are likely to be frustrated if the decisions the organization has already made are effectively off limits for discussion. Participat- ing organizations have scarce resources and will resent being drawn into what they see as empty proceedings. Open procedures are most successful when the host organization leaves itself ample room to adapt as discussions mature. For example, participants may want to see the underlying risk assessment done in a different way, so as to illurn~nate issues of particular salience (e.g., the risks or costs imposed on particular groups, alternative units of measurement). They may call for further data collection to address uncertainties that trouble them most. When participants are asked to contribute to the development of a risk message itself, they may want to explore different strategies for dissemination and additional target audiences. Where time and legal considerations permit, participants may productively help the risk manager to develop new or refine extant risk management options. Once participation has begun, it is important to sustain it. Reg- ular updates, newsletters, and briefings can reinforce the belief that the organization is responsive to input from participating groups. There may be strong disincentives to early efforts at openness. For example, at early stages the organization's risk assessment may be unfinished; openness at this point could result in inconsistent information emanating from different sources within the host orga- nization, which itself could undermine trust and credibility. We do not wish to deny that such complications exist; however, such con- siderations should not be permitted to automatically preclude early participation unless they clearly outweigh its considerable advan- tages. The Empowermerlt Problem Openness is not the same thing as empowerment. Risk man- agers should anticipate some confusion concerning the objectives of
RECOM:MENDATIONS 155 participation. In the past some host organizations have seen par- ticipation as a means to a narrow end the development of better risk messages while outside participants may believe that they have been given a full vote in making the risk management decision (e.g., the choice among regulatory options, the decision to issue ~ public health announcement, the decision to locate a new corporate facility) or in changing the decision process itself. Some participants may feel that discussing risk messages without addressing the risk manage- ment decision itself is beside the point. There is admittedly a fine line between being responsive to outside concerns and relinquishing responsibility to make risk management decisions. It is the risk man- ager's responsibility to be as clear as possible at the outset about where the line is drawn for a particular case. (This does not mean, however, that the risk manager should expect assent on this point, and ambiguity is likely to remain, but it is a matter that is better explicitly discussed than left below the surface.) Outside participants need to understand that, because of statutes and electoral responsibility, an organization cannot, and should not, share its responsibility for risk management decisions. Federal agen- cies, for example, are not commonly able to delegate authority and still remain within their legal authority and thus accountable to the electorate (through executive or legislative oversight) for their reg- ulatory actions. In the past vested interests have been suspected (often by groups that are absent because they cannot afford the costs of participation) of abusing open procedures to thwart or delay de- cisions, rather there to improve them. The host organization, for its part, should not expect participating groups to relinquish their right to raise objections later on, using litigation or other means, simply because they have been consulted in advance. Safeguarding Balance and Accuracy in Risk Messages- Prevent~ng Real and Perceived Distortion For many risk messages, credibility depends on the audience's belief that the message is reasonably objective; there is broad skep- ticism about organizations shading the truth to suit their ends. Because bias, like beauty, is often in the eye of the beholder, it may be very difficult for those who oversee the preparation of risk messages to ascertain, by examining the messages themselves, whether they will mislead audiences or be perceived as distorted. Procedural safeguards may be much more effective.
156 IMPROVING RISK COMMUNICATION To help ensure that risk messages are not distorted and do not appear as distorted, those who manage the generation of risk assessments and risk messages Should: . hold the preparers of messages accountable for detecting and reducing distortion; . consider review by recognized independent exerts of the underlying assesement and, when feasible, the message; . when feasible, subject draft messages to outside preview to determine if audiences detect any overlooked distortions; and . prepare and release a White papers on the risk assessment and risk reduction assessment for comment. Accountability Distortion can enter at two stages: in the preparation of the expert analyses that form the basis of a risk message and in the composition of the message itself. Risk managers should actively encourage those who prepare messages and the expert analysts within the organization to supply materials that are as free of distortion as possible. Risk managers should sensitize employees to particular types of biases and perceived biases that it is particularly concerned about and see that the experts are aware of subtle causes of perceived bias. Experts commonly must synthesize risk information that is fraught with uncertainty, for which many choices among competing quantitative and qualitative assumptions and methodologies must be exercised. To cite but one of the many assumptions that will be found in a particular case, for example, the assessment may be based on "worst case" or on "best estimate" calculations. There is a constant danger that such choices will be unduly influenced by three types of bias: (1) the expert's personal value judgments about what the risk management outcome should be; (2) the expert's belief of where the organization's self-interest lies; and (3) "expert bias," which some- times leads experts to exaggerate the certainty and precision of their assessments. Unfortunately, one cannot assume that experts are sig- nificantly more self-conscious about the subtle distinction between value judgment and scientific consensus in complex analyses than nonexperts are; this means that the risk manager needs to be ac- tively involved in preventing distortion in the way risk assessments and risk reduction assessments are performed and presented.
RECOM:MENDATIONS 157 Some of the assumptions inherent in a risk assessment have to do with uncertainties in the underlying science: for example, choosing among available scientifically supportable theories about extrapolating to humans from animal data on the carcinogenicity of various doses of tested substances. These assumptions unavoidably interject a subjective element into the risk assessment, by reflecting the assessor's judgment about which extrapolation method is most likely to be confirmed by future research. Other assumptions will reflect the values of the scientists per- form~ng the assessment. They may have chosen to use "conservative" estimates in various portions of their work. They may have summa- rized risks to different groups of exposed people in a way that ignores who those people are (rather than, say, giving extra weight to risks to children). Risk analysts and risk managers also may make special assump- tions about how to weight diverse risks. For example, risks that involve horrific outcomes (e.g., cancer deaths as compared to death by cardiovascular disease) and catastrophic outcomes (e.g., release of lethal chemicals in populated areas, as occurred in Bhopal, India) are sometimes given extra weight in making risk management decisions. Such assumptions may be widely accepted value choices. They may be just what the public prefers experts to do when confronting uncertainty. However, they need routinely to be made explicit if audiences are to interpret the resulting risk messages appropriately. Moreover, because they reflect the interjection of values into assess- ment, they need to be cited as a matter of openness and public accountability. More commonly feared by skeptical recipients than expert bias Is the possibility of (intentional or unintentional) distortion to fit ideological precepts. Government organizations are particularly sus- ceptible to suspicions of distortion born of ideological bias. Congress, the press, and advocacy groups frequently charge that agency po- sitions subordinate science to the current administration's ideology (e.g., a preservationist tendency in the Carter administration and a laissez-faire one in the Reagan administration). One should not conclude that such influences are flatly inappropriate in public risk management decisions; we elect presidents and legislators based on their expressed values and platforms and then hold them politically accountable for the value judgments they make when they are in office. Thus, for example, different administrations may properly seek different balances between health risks and econorn~c benefits.
158 IMPROVING RISK COMMUNICATION Similarly, how agency decision makers value usually unmeasurable quantities (e.g., nonmarket goods) in reaching regulatory decisions appropriately depends on philosophy or ideology. However, when the ideology overrides science or blinds the decision maker to established facts, the result is distortion. Risk managers usually rely on information provided to them by staff scientists, engineers, and analysts. Frequently, this informa- tion is generated several levels below the manager and must pass through a series of intermediate managerial and policy reviewers. These reviews can filter out information or positions that are seen to contradict current policies, presenting a danger that the risk manager receives, perhaps unknowingly, distorted, incorrect, or inadequate in- formation. Risk managers should establish an environment in which staff members believe themselves obligated to be honest and to come forward with their best information and analysis, even if it is not entirely welcome. Risk managers should not permit anyone to be pe- naTized for arguing within the organization against the organization's or the administration's position, when the facts point elsewhere. They should remain constantly aware that failure to elicit the best technical information from within the organization can be extremely counterproductive to their credibility. Establishing this environment may be abetted by a formal procedure, such as that established by the Nuclear Regulatory Commission in 1980, known as Differing Profes- sional Opinions (U.S. Nuclear Regulatory Commission, NRC Manual Chapter NRC-4125, September 1980, amended July 1985), and by periodic attention to ensure effectiveness (NUREG-1290, "Differing Professional Opinions: 1987 Special Review Panel," U.S. Nuclear Regulatory Commission, November 1987~. Independent Review To help ensure that choices made in perforrn~ng the risk assess- ment do not introduce errors or analytic assumptions that conflict with areas of current scientific consensus, organizations should rou- tinely subject the underlying assessments, and when feasible the ensuing risk messages themselves, to independent peer review. This review can help managers satisfy themselves that uncertainties are adequately characterized and that scientific disagreements are un- derstood. Peer review should be as independent of the communicating or- ganization as possible and should be conducted by a group whose
RECOMMENDATIONS 159 collective expertise blankets the scientific areas that are germane to the risk message. The Science Advisory Board of the U.S. Envi- ronmental Protection Agency (EPA) and various assessment panels convened by the National Research Council are effective examples. Message Preview When possible, drafts of risk messages and the information on which they are based should be made available to selected outside individuals for their preview and comment. Previews by partisans is a proven method of identifying inten- tional and unintentional slants in risk messages: if value judgments have inappropriately intruded to produce distortion, groups that hold contrary values are certain to proclaim the misstep. In cases where early participation has been possible, the participants themselves can perform the preview. Outside previews will also help reveal where agreement exists among diverse groups. Such coordination can help reduce the inci- dence of needlessly competing or conflicting messages from groups that are in basic accord. There are many instances in which partisan preview is not ad- visable, particularly when it would appear that the organization is unfairly giving advance information on major policy changes to some groups and not to others. (Note: Previews of messages by the general public have also proven effective. "Focus groups" have increasingly been used, less to detect bias and inaccuracy than to judge whether the intended message is actually understood. Although more ex- pensive and time consuming, representative sample surveys can be used to provide a more accurate picture of the likely response of the intended audience.) Written Document The assessment of potential bias, as well as the search for tech- nical errors, is greatly enhanced when written supporting documents are available. When time and resources permit, the communicating organization should synthesize the scientific information base into a formal "white paper" that can be generally released. This document should summarize relevant quantitative and qualitative scientific in- formation, the attendant uncertainty about the risk and about risk reduction alternatives, and the assumptions employed. Federal agen- cies could release such a document as or in conjunction with the
160 IMPROVING RISK COMMUNICATION preamble to a formal notice of proposed rule making, as has been done for major regulations by the Food and Drug Administration and the EPA. Such a document can facilitate the elicitation of reac- tions from review by independent experts, partisan groups, and even the lay public. It can also foster understanding of the risk issue in different parts of the organization itself. Fostering Competence Making Risk Communication Smarter Risk communication has only recently come into focus as a con- cept, and in many organizations it is still subsumed under other functions, such as risk assessment or public affairs. More attention should be paid to risk communication as a distinct undertaking. Successful efforts in risk communication require a blend of technical and communications proficiency in the risk organization. Excluding technical experts can lead to false or incomplete messages or the appearance or reality of the willful manipulation of facts. Excluding those with public affairs functions provides a danger of insensitivity to the capacities, interests, and needs of the audience. Risk managers need to use procedures that attain a balance between two distinct types of expertise: the risk subject matter (e.g., carcinogenic risk, occupational safety) and risk communication. Organizations that communicate about risk should take steps to ensure that the preparation of risk messages becomes a deliberate, specialized undertaking, taking care that in the process they do not sacrifice scientific quality. Such steps include: . deliberately considering the makeup of the intended audience and demonstrating how the choice of media and message reflects an understanding of the audience and its concerns; attracting appropriate communications specialists and train- ~ng t ethnic al staff in communications; . requiring systematic assurance that substantive risk experts within the organization have a voice in producing accurate assess- ments and the derivative risk messages; . establishing a thoughtful program of evaluating the past per- formance of risk communication efforts; and . ensuring that their organizations improve their underst~nd- ing of the roles of intermediaries, particularly media reporters anal editors, including an understanding of the factors that make a risk story newsworthy, of the practical time and space constraints, and of the limited technical background of most media personnel.
RECOMMENDATIONS Assessment of Audience 161 As noted above, a source organization should, before it initiates risk communication, set realistic goals; it should make a deliberate effort to formulate its communication objectives, identify intended audiences, consider alternative communication strategies, and assess the likely usefulness of a message to the audience. Risk communication cannot be considered an informal add-on to the technical assessment effort. Effective risk communication involves specialized knowledge of, and when feasible interaction with, the intended target audience, an understanding of media practices, and an appreciation of the role of other intermediaries in relaying and translating messages. Those who assess risks and risk control options within an organization are not usually experienced in these areas. As soon as the organization's risk communication objective is es- tablished, analysis of and interaction with the target community, or its representatives, should commence; deliberate audience research is important. Ideally, an audience profile should be compiled that describes the nature of the members of the audience and gives some idea of whom they trust, what they believe, and what concerns and worries motivate their actions; focus groups, surveys using represen- tative sample techniques, and demographic studies may be helpful in compiling the profile. Available time and resources do not always permit the compilation of a detailed profile, but the risk manager should realize that risk communication will suffer to the extent that the audience is mischaracterized. The results of such audience re- search should be made public in a timely manner. Failure to do so may undermine the apparent openness of the organization. Specific knowledge of the intended target groups permits intel- ligent segmentation of the audience, another key to effective com- munication. A uniform message will have varying effects on differ- ent individuals. Audience segmentation is useful both to customize the message and in the choice of the most effective communication channels. Risk communicators need to be aware that individuals may prefer to use different channels for different aspects of their decision-making process. Some channels, for example, may be best for conveying general knowledge but less reliable for affecting whether individuals believe a risk is or is not something to be worried about. Risk managers should expect those who prepare risk messages to construct a communication plan that clearly links the choice of channel and customized messages to an understanding of audience
162 IMPROVING RISK COMMUNICATION segments and that links the definition of audience segments to an objective understanding of the audience population. Two caveats should be noted. First, while an explicit analysis of the audience is important, it should not be expected to supplant all other considerations in planning for the risk message. Other important organizational goals (legal constraints, consistency of cur- rent and past policy, support of current enforcement efforts) must be factored in. The risk manager's difficult job is to attain a rea- sonable balance among these competing organizational objectives. Second, but no less important, the risk manager must be concerned with the outside appearance of the explicit communication planning effort. Observers may rightly or wrongly perceive a deliberate ef- fort to understand, segment, and reach the audience as inherently manipulative and invasive. Specialized Talent Risk communication requires specialized knowledge and talent. It may be difficult to adequately reeducate technical or other existing staff to coordinate the message preparation effort. Preparing and helping implement the explicit risk communication plan described in the previous section require special expertise. Specialized knowledge in such subjects as demographic techniques, the psychology of risk perception, and how the media work, combined with the rare knack for writing clearly about complicated technical issues, is needed. Recruiting staff with such capabilities or retraining existing staff amounts to putting the task of risk communication on a pro- fessional level in the organization in order to achieve better-informed risk communication decisions. However, skeptics inside and outside the organization may see it as importing dubious strategies and tech- niques from marketing and advertising into the heretofore scientific domain of risk assessment. Vigilance must be applied (open proce- dures can be of great value here) to ensure that such techniques do not become manipulative or deceptive in fact or appearance. Scientific/l~echnicat Accuracy and Completeness Upgrading the staff that coordinates the preparation of risk messages to a professional level does not mean that substantive experts within the communicating organization can be shunted away from the process. They must remain involved in order to ensure that factual errors are not introduced.
RECOMMENDATIONS 163 Technical inaccuracy and incompleteness in message content can easily be used by knowledgeable advocates of alternative positions. For example, national advocacy groups, including consumer and envi- ronmentaTist groups, use competent scientific/technical professionals in presenting positions and in countering government antic corporate press releases. Once a message has been shown to be inaccurate or misleading, organizational credibility is lost for that message and for succeeding messages on quite different topics. A technically flawed risk message may reflect poor risk commu- nication within the communicating organization. To prevent this, risk managers should require that senior technical staff have an op- portunity to evaluate the quantitative and qualitative accuracy of risk messages and that any exceptions are clearly reported. Evaluation and Feedback Even when communications professionals help design and guide the risk communication effort, doubt will remain about whether and how the intended audience will apprehend the message. Source organizations should routinely conduct retrospective eval- uations of their communication efforts and of particular messages. At this stage there appears to have been remarkably little formal eval- uation by organizations that communicate about risks. Evaluation, if coupled to a feedback mechanism, is a necessary step in ensuring improvement in the competence of an organization's risk communica- tion program. Organizations that disseminate risk messages should institute formal programs that assess their experience. Evaluations should address both questions of content and questions of process, as described in this report. That is, the effectiveness of messages should be examined along with a sense of how different channels and intermediaries affected transmission- but attention also needs to be devoted to the organization's performance with its procedures for setting realistic goals, involving interested and affected parties, attaining balance, and creating internal expertise in risk communi- cation. Role of Intermediaries Most risk messages will pass through one or more organizations or individuals before reaching the final recipients. Sometimes the only way of ensuring that a message reaches the people for whom it
164 IMPROVING RISK COMMUNICATION is intended is to rely on intermediaries. In any case an organization initiating risk communication should identify the intermediaries who will handle its messages, assess their needs and constraints, and adjust to those conditions if possible. Journalists look first for clear statements about events and issues at conflict. They operate under strict deadlines and compete for allo- cations of space or time. Providing journalists with written copy will reduce, but not eliminate, the chances of being misquoted. Regular contact with journalists, including after stories have appeared, will generally improve the basis for later exchanges. Community organizations and prominent individuals can be ef- fective intermediaries for risk messages. But health departments, public libraries, professional associations, voluntary organizations, and similar groups all have their own aims ant! purposes. Discov- ering which organizations or individuals would be appropriate in a particular situation and developing the working relationship that is necessary for constructive interaction with such intermediaries can require considerable time and effort. It can, however, make the difference in reaching the intended recipients. Some Notes on Handing Risk Communication In Crisis Conditions Many risk situations require that risk messages be delivered im- mediately: examples include emergency conditions, challenges to an organization's positions before the organization is prepared to re- spond, and intense and contentious public controversy. In that atmo- sphere the deliberate procedures recommended above (e.g., outside reviews and analysis of the audience) may well be impractical. The process for risk communication in crisis conditions requires special care. Risk managers should ensure that: . where there is a foreseeable potential for emergency, advance plans for communication are drafted. These plans should be draftee! jointly with the intended audiences (e.g., local communities near a chemical plant, paramedics, and fire departments). Such plans should be prepared in the context of concrete events and scenarios, should provide specific information that is relevant to people's risk- averting actions, and should specify actions that may be taken in case of a disaster or emergency; and . there is provision for coordinating information among the various authorities that might be involved and, to the extent feasible,
RECOMMENDATIONS 165 a single place where the public and the media can obtain authoritative and current information. THE CONTENT OF RISE MESSAGES . ~ The preceding section is addressed to risk managers, who have overall responsibility within their organization for assessing risks, making risk management decisions, and managing risk communica- tions. This section is addressed to those within the communicating organization who are responsible for preparing formal risk messages. In general, we find that practical advice on the content of risk messages depends heavily on the particular situation; for example, a public health advisory message on AIDS and an EPA announce- ment on the regulation of the use of a pesticide for certain crops may have quite different purposes, audiences, urgency, and visibility. We concentrate here on four generic matters audience awareness, uncertainty, comparative risk, and completeness that have been the source of difficulty in the past over a broad range of risk communi- cation efforts. Relating the Message to the Audiences' Perspectives Risk messages are often based on the information in special analyses prepared for internal organizational purposes (e.g., to assess whether a particular risk exists or what risk management option to choose). That information often reflects the prior knowledge, perspectives, and language of risk experts and risk managers. It may not be sufficient for effective risk messages. Risk messages should closely reflect the perspective, technical capacity, and concerns of the target audiences. A message should: . emphasize information relevant to any practical actions that individuals can take; be couched in clear and plain language; respect the audience and its concerns; and . seek strictly to inform the recipient, unless conditions clearly warrant the use of ~nduencing techniques. Personal Relevance Consideration of the specific decisions that recipients face pro- vides the surest basis for determining what risk information to em- phasize in a risk communication. Such decisions might be whether
166 IMPROVING RISK COMMUNICATION and how to change personal behavior to respond to a reported health risk, whether to use or avoid a product that is being regulated, how to vote on a local siting issue, and whether to follow a particular risk issue further. Much of the information available to those who prepare formal risk messages has been assembled in risk assessments prepared in a context of risk management decision making. The basic question in such assessments is, "What should the organization do (if anything) to reduce risk to the population?" Estimates of total exposures, to- tal risk reduction costs, and other aggregate data often written by experts whose immediate objective, understandably, is to make them scientifically defensible in the eyes of other experts often predom- inate. The central question answered in a risk message should be "What should the recipient know to improve the choice among per- sonal options (including the consequences of doing nothing)?" Data and analyses that risk experts have not emphasized may be needed. In the terms of decision theory, a risk message should contain infor- mation to which those decisions are "sensitive" the facts that are most central to the choice at hand. This criterion should determine the kinds of information included and the detail and precision with which it is presented. For some decisions the critical information is the magnitude of the risk involved; for others it is the processes by which risks are created and controlled. Risk information should be expressed in terms of risk to a rep- resentative individual, not only as a general population estimate. If there are highly exposed or particularly sensitive subgroups, such groups should be identified in a way that individuals con understand if they have reason for concern. Practical advice on such matters as danger signals of exposure, available remedies, sources of help, and so on should be included. Selecting information relevant to individual choice is particularly important for risk messages-health warnings are prime examples- that are intended for an audience that is not already motivated to listen. The existence of such risks may mean little if it is not made clear what practical measures an exposed individual might use to avoid or reduce them. Clarity The risk message should be understandable to the target au- dience. When there is doubt about the ability of the audience to
RECOMMENDATIONS 167 absorb technical material, little is lost by assuming that the audience has little technical training. Carefully chosen, vivid, concrete images and the use of personalized examples can help a lay audience to un- derstand and can often even ensure understanding among those who are more familiar with the subject risk. Message designers should try to avoid using images or terms ("morbidity" Is one example) that, while seemingly familiar to laypersons, have different or more precise technical meanings for experts. Special care is needed in depicting statistical concepts and prob- abilities. Few people can meaningfully distinguish among small prob- abilities ant] may have no way of determining if such an assessment as "l-in-10,000 lifetime risk" is worth worrying about. In a Tong message with extensive technical detail or quantitative complexity, the key portions conclusions, summary, and recom- mended actions should be written in lay terms. The pursuit of clarity is likely to be enhanced by experimen- tation, post hoc evaluation, and the pretesting of messages with laypersons, all of which have been discussed above. In our view clarity is a necessary but not a sufficient condition for improved risk messages. This is because while such features as plain language and vivid examples can enhance understanding, they can also, if misused, potently enhance misunderstanding. One pit- fall is that of equating clarity with brevity. The message preparer's goal should not be to gloss over the complexity and uncertainty of a risk but to reflect those qualities in plain language. Those who pre- pare risk messages should expect that their attempts at brevity will provoke protest among those who fear it will lead to greater m~sun- derstanding. Where the nature of the chosen communication channel requires a short message, as with mass media announcements, this of course poses an unavoidably difficult dilemma and one in which the procedural measures recommended above (e.g., openness and message pretesting) may be vital. Respect for the Audience and Its Concerns If a message appears insensitive to an audience's actual concerns' there is a real chance that the audience will be alienated. The mes- sage should not disparage people's subjective reactions as inferior to expert assessments. If members of the audience hold beliefs that the source organization sees as false, it is better for the message to address these beliefs than to omit them as irrelevant. If people
168 IMPROVING RISK COMMUNICATION are advocating specific options for reducing risk, the message should address them, even if the source organization's analysis shows the options to be infeasible technically or legally. The impact of a mes- sage is, of course, determined as much by style and genera] demeanor as literal message content. In all cases, but particularly in face-to-face delivery of messages, care should be taken to show compassion and to avoid distant, anti- septically statistical treatment of illnesses, injuries, and cleath. For example, if a person is gravely concerned about a particular hazard, a message that dismisses the risk as trivially small will surely come across as coldly patronizing. The best way to summarize this general point is to observe that legitimacy is inherently reciprocal in nature; only if a source acknowledges the legitimacy of the audience's felt concerns will it have a chance to be seen as legitimate itself. Use of ~Inpuence Stralregies~ Those who prepare risk messages, and particularly those in gov- ernment organizations, need to be circumspect about using "influence strategies" in their risk messages to influence recipients' beliefs or ac- tions, and they should expect their audiences to suspect attempts to influence even when the intent is simply to inform. Americans are usually most comfortable with risk messages that, in Jefferson's words, "inform their discretion," but that do not at- tempt to advise them how to act in response. Some would draw a line for risk messages at the function of describing the risks and other outcomes (e.g., costs) associated with alternative risk management options, claiming that to go further involves the application of value judgments that are beyond the proper reach of the message source. Others would point out that governments make such judgments com- monly, as when they by law or regulation establish sanctions against certain private actions (e.g., polluting, littering, and not wearing seat belts). In addition, Americans want their public servants to be strongly committed to the pursuit of their agency's national mission, and such individuals understandably form strong views on what they see as the correct ways to respond to problems. They will want to express those strong views. Not understanding many of the concepts presented in this report, these public servants may see a dilemma with respect to their role as risk communicators.
RECOMMENDATIONS 169 In Chapter 4 we introduced a distinction between two risk mes- sage strategies, informing and influencing. Influence strategies com- prise a range of techniques ranging from, at one end, messages that attempt to "persuade" through the selective use of factual informa- tion to outright deception at the other extreme. The user's intent is to convince a recipient to accept the source's opinions or pre- scribed actions. The audience response to a message that is seen as using influence strategies may be to count it as illegitimate. The audience may disregard the entire risk message as slanted toward a predetermined outcome. Credibility is a casualty. When is an influence strategy appropriate? As we noted in the discussion in Chapter 4, Americans accept influence strategies in some settings. Dietary warnings by public health officials are an example of the generally accepted use of influence. There is some indication that reigning traditions vary according to the culture of the professional field of the risk assessors traditions in dam safety, for example, may differ from those in toxicology with respect to practitioners' efforts to prescribe policy or personal choices. Risk managers in government agencies would be well served to know when the use of influence strategies is safely legitimate and when it is not. In general, we would urge great care in the use of influence strategies by government agencies. We have identified three partic- ular situations in which use of influence strategies by government may arouse resentment that could affect the credibility of a message and/or source: i. When there is unresolved public controversy over the issue, particularly if there has been no public forum at which relevant voices have had their say. Whenever government attempts to influ- ence citizens' beliefs and actions, it should be able to point to some legitimate public process one that has given interested and affected parties a chance to express themselves which concluded that us- ing risk messages to influence behavior serves an important public purpose. 2. When the form of influence strategy is toward the more severe end of the spectrum of influence techniques (i.e., near deception). 3. When there is no evident threat of externalized effects that is, when the risk is confined largely to the persons who themselves undertake the risky behavior, without endangering others. When influence strategies are used, risk messages should at- tempt to distinguish the analytic function of describing risk from the
170 IMPROVING RISK COMMUNICATION prescriptive function of advising recipients about what to do. It is an error to imply that the technical analysis led irrefutably to the prescription. The organization that disseminates the message should make clear that, in balancing risks and other factors to arrive at its recommended action, it has made a policy judgment. It is, of course, politically accountable for such judgments. Handling Uncertainty Uncertainty is a central fact in the assessment of many contem- porary risks. It is usually present both in risk assessment and in the assessment of risk management options. The way that risk messages treat this uncertainty can have a major influence on the effective- ness and credibility of a communication effort. A major difficulty is avoiding unnecessary confusion between scientific uncertainty on one hand and policy disagreement on the appropriate risk management approach on the other hand. Risk messages and supporting materials should not minimize the existence of uncertainty. Data gaps and areas of significant disagreement among experts should be disclosed. Some indication of the level of confidence of estimates and the significance of scientific uncertainty should be conveyed. There are dangers if existing uncertainty is widely perceived as either underplayed or exaggerated. Any attempt to minimize uncertainty may make it appear that the caveats expressed by experts are being ignored. Exaggerating uncertainties can have the effect of obscuring the scientific basis of a risk management decision (e.g., whether to regulate, whether to issue a health advisory to the public), leaving the audience with the impression that the decision has been arbitrary in nature. One reason the effectiveness of risk messages is so sensitive to their treatment of uncertainty is that the handling of uncertainty is a central issue in many of today's risk controversies. Often one side in a controversy will emphasize the need to base important risk management decisions on sound science, rather than on mere conjecture. As often, the other view will emphasize that ordinary prudence "better safe than sorry" dictates that action can be taken before conclusive scientific proof comes in. A central dispute thus becomes, "How much proof is needed?" and the degree of extant scientific proof itself becomes a matter of close partisan scrutiny.
RECOMMENDATIONS 171 Those who prepare risk messages commonly must choose be- tween presenting the full range of available estunates, presenting a restricted set, or offering a single estimate based on consensus among consulted experts. Choosing any of these methods has its dangers, and more complicated presentations of the range of uncertainty are often needed; what is appropriate depends on the data available for a particular case. It is usually dangerous for messages to characterize the overall level of uncertainty quantitatively, as might be done by describing statistical confidence intervals. In most situations expert assessments have multiple sources of uncertainty, and statistical measures do not adequately represent the complexity of the analysis. For many messages an extensive description of uncertainty obvi- ously cannot be included in the text itself. However, it remains useful to have prepared an explicit account, even if for practical reasons it must be consigned to supplementary documents made available to recipients upon request. In general, those preparing risk messages are best served if they have available to them a statement of the scientific conclusions of the assessment of a professional quality that might be used for materials intended for expert peer review, such as papers submitted to profes- sional journals; this will help ensure that uncertainties and necessary qualifications are adequately conveyed from the experts to those who prepare messages. A form of sensitivity analysis can be helpful. To gauge the signifi- cance of uncertainty and of differences among experts, it is frequently helpful to vary the different sets of expert estimates systematically and then to gauge the effects on the overall risk estimate. The esti- mate will be more sensitive to some choices of assumptions than to others. If the risk message uses one of the competing assumptions, the risk message should say so, disclose why it was chosen over oth- ers, and indicate what difference it makes to the assessed risk. It should be observed that this procedure is one for which the needs of risk communication may dictate how a risk assessment itself is done. The general goal of this recommendation is to help audiences dis- tinguish areas of scientific agreement amid what may appear as vast areas of policy disagreement. The advantage of careful delineation of existing scientific uncertainty is that it gives audiences a sense of the degree of scientific consensus and allows them to distinguish minor from major uncertainties.
172 . IMPROVING RISK COM:MUNICATION Comparing Risks One factor that inhibits public understanding of risk messages is that people often cannot easily relate the low say, 1 in 10,000- risk probabilities presented to their everyday experience. They are thus often deprived of a sense of the personal meaning of the risk in question and so cannot arrive at a comfortable decision about whether to take actions to deal with the risk or whether to be concerned at all about the hazard. In theory, at least, this difficulty can be overcome by quantitative comparisons between risks between familiar and less familiar risks). Risk comparisons can be helpful, but they should be presented with caution. Risk comparisons must be seen as one of several inputs to risk decisions, not as determinants of decisions. There are proven pitfalls when risk e of diverse character are compared, especially when the intent of the comparison can be seen as that of minimizing a risk (by equating it to a seemingly trivial risk). More usefi~} are comparisons of risks that help convey the magnitude of a particular risk estimate, that occur in the same decision context (e.g., risks Tom dying and driving to a given destination), and that have a similar outcome. Multiple comparisons may avoid some of the worst pitfalls. More work needs to be done to develop constructive and helpful form of risk comparison. In theory, at least, comparative information should be an attrac- tive element of risk messages. We have advised that the best risk messages are those that inform the recipient's actual choices, and increasingly those choices are between courses of action (or inaction) that represent different risks. Risk comparisons ideally might help individuals steer a prudent course between risks of various sizes. However, actual attempts to compare risks have engendered con- siderable controversy and distrust. One reason for this is the fear that comparisons will be used to influence and even mislead the lay public. Individuals are known, for example, to subjectively underes- timate actual incidence rates for some fatal risks (e.g., those resulting from asthma and strokes) and to overestimate others (e.g., risks that are especially feared, like those resulting from tornadoes and bo- tulism). Thus, comparing a risk to the likelihood of death by asthma would probably induce most people to similarly underestimate it. Another difficulty is that alternatives often have more than one risk attribute, and different people emphasize different facets. For a particular choice, for example, one group might concentrate on the relative number of deaths associated with each alternative, and a
RECOMMENDATIONS 173 second group may emphasize the way risks (or costs) are distributed among various groups within society. The choice of any single metric for comparison will thus ignore facts that some observers may value highly. Some who have used comparative risk information seem to have done so on the assumption that recipients would, upon seeing that a particular risk is small, elect to stop worrying about it. Implicit was the notion of an "action threshold," or perhaps a "worry threshold," that would render subthreshold risks unfit for serious consideration. The notion has a debilitating flaw, and it is not surprising that risk comparisons that seemed to be used to trivialize certain risks met with objections. Personal and organizational risk management decisions are based on many factors, of which a risk estimate is only one. For example, even a trivial risk may be worth eliminating if the costs of elimination are negligible; to suggest that people should decide based on one factor for example, expected mortality alone is somewhat analogous to saying people should make purchases based solely on comparative pricing without considering the value of the product to them. In practice, risk comparison data can rarely be closely linked to specific decisions in the absence of other critical information about decision options. In general, comparisons of "unlike" risks should be avoided, as they have often either confused message recipients or irritated them because they were seen as unfair or manipulative. Directly comparing voluntary (e.g., skiing) and involuntary (e.g., air pollutants) risks, or natural (e.g., earthquakes) and technological (e.g., food additives) risks, for example, is rarely a good idea. More generally, those who prepare risk messages should appreciate the weakness of risk comparisons as a means of placating people about risks that are calculated to be small. When can comparisons be used in a risk message? Three situa- tions suggest themselves: 1. To help message recipients comprehend probabilities. In iso- lation a term like "one chance in a million per year" may convey little. An analogy to lengths (1 inch to 16 miles) or volumes (1 drop to 16 gallons) may help some people; reference to other known one- to-a-million risks of the type under discussion (for lung cancer, that of smoking a certain number of cigarettes; for private transportation mortality, that of traveling 300 miles by car) may help others, if they have a grasp of the reference risk.
174 IMPROVING RISK COMMUNICATION 2. To directly compare alternative options. Personal and orga- nizational decisions can be better informed if the risks of alternative actions are bald out in comparable terms. Comparing the risks of coffee and tea consumption, or the risks of air and automobile travel between two points, may improve one's ability to make informed choices (again, however, one would not expect a risk comparison to necessarily dominate in such choices). For a regulatory agency the health risk of a pesticide may be directly compared to that of its substitute if it were removed from commerce. 3. To gauge the relative importance of different causes of the same hazard. Discussions of public and private actions with respect to indoor radon may be improved, for example, by a comparison of radon with smoking and other known causes of Jung cancer. One interesting approach is the use of risk ladders, for which a range of probabilities is presented for a single class of risks. The discussion of Figure 5.1 shows the limitations of past use. If one is careful, however, the use of multiple comparisons helps counteract the possibility that people may severely rn~sestimate a particular risk, even though it is farn~liar to them. It also reduces the danger of arousing the scientific disputes that can often arise when only two risk estimates are compared, one or both of which are subject to scientific debate. Ensuring Completeness If the information in a risk message is incomplete, the recipients may be unable to make well-informed decisions. A complete information base contains five types of qualitative and/or quantitative information: (1) the nature of the risk, (2) the nature of the benefits that might be affected if risk were reduced, (3) the available alternatives, (4) uncertainty In knowledge about risks and benefits, and (5) management issues. There are major advantages in putting the information base into written form as an adjunct to the rislr message. Those who prepare risk messages should ensure that the mes- sages are complete. A suggested risk information checklist of relevant topics for the design of a complete message, drawn from the descrip- tion in Chapter 2, is summarized in Figure 7.1. Two points are worth emphasis. First, a complete risk message, as we have defined it, includes information other than a risk assess- ment; it covers the characterization of current or possible efforts to
RECOMMENDATIONS INFORMATION ABOUT THE NATURE OF RISKS 1. What are the hazards of concern? What is the probability of exposure to each hazard? What is the distribution of exposure? What is the probability of each type of harm from a given exposure to each hazard? What are the sensitivities of different populations to each hazard? How do exposures interact with exposures to other hazards? What are the qualities of the hazard? 8. What is the total population risk? INFORMATION ABOUT THE NATURE OF BENEFITS 1. What are the benefits associated with the hazard? 2. What is the probability that the projected benefit will actually follow the activity in question? 3. What are the qualities of the benefits? 4. Who benefits and in what ways? 5. 6. 7. How many people benefit and how long do benefits last? Which groups get a disproportionate share of the benefits? What is the total benefit? I NFORMATION ON ALTERNATIVES 1. What are the alternatives to the hazard in question? 2. What is the effectiveness of each alternative? 3. What are the risks and benefits of alternative actions and of not acting? 4. What are the costs and benefits of each alternative and how are they distributed? UNCERTAINTIES IN KNOWLEDGE ABOUT RISKS 1. What are the weaknesses of available data? 2. What are the assumptions on which estimates are based? 3. How sensitive are the estimates to changes in assumptions? 4. How sensitive is the decision to changes in the estimates? 5. What other risk and risk control assessments have been made and why are they different from those now being offered? INFORMATION ON MANAGEMENT 1. Who is responsible for the decision? 2. What issues have legal importance? 3. What constrains the decision? 4. What resources are available? FIGURE 7.1 Risk message checklist. 175 reduce risk. Some topics include the cost of control, who pays, how effective the approach is, and whether the control implies additional risks of its own. Uncertainty in the analysis of risk control measures should be included. The message should also contain pertinent in- formation about how any risk management decision has been or will be made.
176 IMPROVING RISK COMMUNICATION Second, the checklist used for preparing a complete risk mes- sage should be used to ensure that the underlying analysis itself is complete; that is, concern for risk communication should influence the conduct of risk assessment and risk control assessment. If the information base developed in the analytic process is incomplete, the risk message will be deficient. There are advantages to compiling and keeping the information base in written form. In at least some cases, for example, it will prove useful to compile a "white paper" of factual information on the subject risk. As described in the section above on management of the process, a written record provides a useful management tool for risk communication; if the underlying information is in written form, it can be examined (and perhaps improved) by others inside and outside the organization, helping to prevent surprises when the risk message is disseminated. Such a document also can provide a useful single source for diverse messages, enhancing consistency and accuracy. When feasible, this document should be made available as an adjunct to the formal risk message. Whether or not the information base is compiled in written form, risk communicators should treat it and be seen as treating it as work in progress that is continually subject to improvement. Discussions and debates that surround a risk message often raise new questions, and new data can arise from research and other sources. A CONSUMER'S GUIDE TO RISE AND RISE COMMUNICATION A major theme of this report is that risk communication should be understood to be a two-way interchange between source organiza- tions and those, including the public and its representatives, who are the intended recipients of risk messages. In the previous pages we have directed many recommendations about the process and content of risk communications efforts to source organizations, specifically government agencies and large corporations. If risk communication is a two-way enterprise, both sides have rights and responsibilities that must be understood if the process is to work well. The following recommendation is directed at improving the recipient's ability to participate meaningfully in risk management and risk communication. It is based on the conclusion that, at this stage, nonexpert participants have different understandings of the nature of risk and how it is managed. It is also based on the
RECOMMENDATIONS 177 conclusion that the risk communication process would benefit if the interested public were better able to ask intelligent, probing questions of those in government, industry, and elsewhere who prepare risk messages for their consumption. As source organizations become more accomplished at risk communication, we expect that there will be more opportunities for two-way interactions. We believe there needs to be a national locus for improving the public's ability to participate. Major government and private organizations, including envi- ronmental and consumer groups, that sustain risk communication efforts should jointly fund the development of a Consumer's Guide to Risk and Risk Communication. The purposes of this guide would be to articulate key terms, concepts, and trade-offe in risk communi- cation and risk management for the lay audience, to make audiences better able to discern misleading and incomplete information, and to facilitate the needed general participation in risk issues. Such a guide should: . involve support from, but not control by, the federal govern- ment and other sources of risk messages; . be under the editorial control of a group that is clearly ori- ented toward the recipients of risk messages, and under a~ninistra- tive management by an organization that is known for its indepen- dence and familiarity with lay perspectives, and that can undertake the needed outreach and public information effort; and . cover subjects such as those suggested below e.g., the nature of risk communication, the concepts of zero risk and comparative risk, and evaluating risk messages and others designated by project participants. We believe that the development of such a guide would have several advantages. It would help orient the interested public and the leaders of organized groups and prevent some of the misunder- standing that has occurred in the past. It would provide nonexpert participants with tools and concepts to enhance their participation, including sections about how to identify incomplete, imbalanced, or misleading messages. The process of writing it would advance na- tional discussion about areas of current controversy among players in an often adversarial process of making risk management decisions. The guide would also articulate the basis for public skepticism that sometimes causes consternation among those responsible for risk management and the design of risk messages.
178 IMPROVING RISK COMMUNICATION Project Support It is important that major risk communicators federal agencies, large corporations" support the project. We would expect that the project would require about ~ year to complete and that it would require a full-time staff of two or three persons. Allowance should be made for wide distribution of gratis copies of the final document. Provision should be made to update the guide 3 to 5 years after it is published; updating will help ensure that there is a national focal point for continuing interactions among the groups that, together, can bring about long-term improvement in risk communication. Project Management Editorial control of the guide should be exerted by a steering group in which the views and concerns of the lay recipients of risk messages are paramount. It should not be difficult to identify indi- viduals who reflect an appropriately broad range of lay perspectives. The steering group should also include a minority of other relevant perspectives (e.g., risk managers, scientists and other experts, media representatives, and advocacy groups). The project requires a stable but independent administrative home. For practical reasons it would be most suitably placed under the aegis of an existing organization in order to permit an efficient start-up and a reliable dissemination/outreach phase. The admin- istrative home should be one that is credible to all sides involved in risk management issues and one that has demonstrable relevant experience. The League of Women Voters and the National Safety Council are two of several organizations that meet these criteria. An integral part of the project should be the design of a dis- semination effort that, among other possibilities, makes use of com- patible existing efforts at public outreach involving aspects of risk by professional (e.g., American Bar Association, American Medical Association, American Chemical Society) and other groups. Content of the Gliide We offer a brief topic list as representative of subjects to be cov- ered in a consumer's guide (see Figure 7.23. In addition to coverage of these points and other subjects raised during the guide project itself the guide might contain a directory of information resources on risk topics for the lay public and groups that represent it.
RECOMMENDATIONS 179 WHAT IS RISK? Key Terminology and Concepts Hazard, exposure, probability, sensitivity, individual risk, population risk, distribution of risk, unattainability of zero risk Qualitative Attributes Voluntariness, catastrophic potential, dreadedness, lethality, controllability, familiarity, latency WHAT DOES RISK ASSESSMENT CONTRIBUTE? Quantification Quality, completeness, uncertainty, confidence Scientific and Policy Inferences Assumptions, assessment of benefits, risk management choices WHAT IS THE ROLE OF THE RISK COMMUNICATION PROCESS? Setting Public debate about decisions, informing or influencing personal action Purpose Messages can inform, influence, or deceive Interaction Among Participants Contending conclusions, justifications, credibility, and records HOW CAN YOU FIND OUT WHAT YOU NEED TO KNOW? Technical Content Demystifying jargon, comparing relevant risks, finding trusted interpreters Independent Sources Information clearinghouses, academic or public service sources HOW CAN YOU PARTICIPATE EFFECTIVELY? Finding the Right Arena Identifying the responsible decision maker, getting on the agenda Intervention Identifying points and times for intervention, marshalling support HOW CAN YOU EVALUATE THE MESSAGES AND THE COMMUNICATORS? Accuracy Factual base, track record, consistency, self-serving framing, use of influence techniques, misleading risk comparisons Legitimacy Standing, access, review, due process justification Interpreting Advocacy Comparing competing arguments, seeing where information has been omitted questioning message sources . FIGURE 7.2 A consumer's guide to risk and risk communication. RESEARCH NEEDS As a result of our deliberations, we recommend the nine specific research topics listed below. Some stem directly from the problems identified in Chapter 6. Others are based on our review of available information and the substantial practical experience of committee members. Two criteria guided our selection of topics: (1) additional
180 IMPROVING RISK COMMUNICATION knowledge would lead to material improvement in risk communica- tion practices and (2) creation of such knowledge is likely, given past results and current research methods. We have not set priorities among the topics. Risk Comparison If performed thoughtfully, risk comparison holds promise of mak- ing risk communication more relevant and meaningful to recipients. However, three issues need to be explored to prevent past shortcom- ings of the technique. . Comparability. When are two risks "similar" enough in na- ture to be compared without misreading, confusing, or angering re- cipients? What are the crucial dimensions across which risks should not be compared? . Apprehension of risk magnitudes. How do people apprehend the magnitudes of risks; in particular, how do they interpret very small probabilities, which often seem beyond most people's intuitive understanding? How do different ways of presenting risk magnitudes affect people's feeling for the size of risks? Validation. The use of risk comparisons is undermined if there is doubt about the validity of the data that are compared. Risk estimates used in risk comparisons must be validated in two ways: (1) as to the current scientific accuracy and the associated uncer- tainty or qualifications and (2) as to whether nonexperts are known to systematically underestimate or overestimate such estimates sub- jectively (which would make them inappropriate as "anchors" in risk comparisons). Risk Characterization We need better ways of presenting complex information about risk clearly and accurately and better understanding of the limita- tions of techniques for simplifying complex material. How do people respond to alternative ways of characterizing risks, including alter- native treatments of uncertainty? Role of Message Intermediaries We need a better empirical base for understanding the role of intermediaries in carrying and translating risk messages. What chan- nels (mass media, specialized media, advocacy groups, community
RECOMMENDATIONS 181 organizations, localprofessionals and other opinion leaders, casual acquaintances) do people actually use? How do people validate and integrate messages from multiple sources in deciding what to do, or what to believe, about a particular risk? Case examinations and a review of research in allied fields (e.g., medical education) can help elucidate the direct and indirect flow of information from source to recipient. Pertinency and Sufflciency of Rim Formation Risk communicators need to focus on the information that is most pertinent to recipients' needs; they are in danger of wasting the limited access they have to their audience if they are viewed as preoccupied with marginal issues. What types of information do people actually find pertinent in reaching personal decisions about risk? How does this compare with what the risk manager or decision analyst thinks should be pertinent? How and when do people deter- mine that they do not need additional information in order to decide what they will do about a risk? What information appears necessary to trigger active personal concern about a risk? Psychological Stress Given the number and variety of known risks in modern life, what conditions are necessary to induce stress about a particular risk in persons and communities? Which of the messages that appeal to fear, or that advert to imminent danger, actually cause stress? If people are stressed about a particular risk, how is their apprehension of risk information affected? Recipiente"'Mental Modeled . The information in risk messages is useful only if recipients can Incorporate it into their prior thinking about the risk and its man- agement. Only by better knowing how recipients conceptualize risks and their risk decisions can people create more effective messages. In particular: . How do people think about the risk decisions that confront them? For example, what alternatives do they consider, and what consequences are they aware of?
182 IMPROVING RISK COMMUNICATION ~ How do people think about the causal processes that create risks? For example, do they misconceive exposure processes, and how effective do control efforts seem, intuitively? ~ How do people perceive the social/governmental processes involved in managing risks? For example, what do they believe reg- ulatory agencies are empowered to do, and when are public interest advocates seen as credible? Risk Literacy How do people learn the "analytic" concepts and language they need to understand risk statements? Do they lack important con- cepts? What kinds of materials, including special curricular materials in science and mathematics education, might be effective? Retrospective Cases There is a dearth of case studies that focus directly on risk communication. In particular, retrospective case materials should be prepared that: . Examine risk communication processes, including such topics as the role of experts and others in message preparation, whether and how outside groups were involved in risk management and risk communication decisions, and the role of intermediaries in message · e transmission. . Analyze the responses of recipients and how the responses corresponded to the expectations of the source. Contemporaneous Assessments of Risk Cases Too seldom are there attempts to learn from ongoing cases of risk management. This is partly due to an understandable desire to concentrate resources on solving a risk problem, rather than cal- ibrating it; nonetheless, real-time assessments can provide valuable knowledge for making general improvements in risk communication. This contemporaneous research should address such matters as how people react to different types of messages and channels; what their actual concerns, frustrations, and data needs are; and how effective alternative communication and message strategies are.