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.
4 Purposes of Risk Communication and Risk Messages In this chapter we distinguish two types of settings public de- bate and personal action in which risk decisions and risk communi- cation occur, and we show how the risk communication process and its participants vary in these settings. We then discuss two distinct purposes of risk messages informing and influencing-that coexist in risk communication, sometimes even in a single risk message. Fi- nally, we address the thorny ethical problem of the appropriateness of influencing as a purpose of risk messages, particularly messages that public agencies distribute to citizens. SETTINGS OF RISE COMMUNICATION Public Debate In a setting of public debate such as congressional hearings, congressional debates, formal regulatory adjudication, and notice- and-comment rule making democratic risk communication includes a wide range of messages, sources, and audiences. Interested groups raise questions for the experts, who respond; experts from different perspectives dispute with each other; and citizens and their represen- tatives dispute using, among other things, the experts' findings and criticisms of each other's results. Messages describing and surnrnariz- ing scientific knowledge about risks and benefits are important, as are 72
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 73 critiques of those messages and that knowledge. In the United States, regulatory decisions must generally be based on the best available scientific knowledge to be defensible against legal challenges. As a result, much risk communication in the regulatory context clears with the adequacy and proper interpretation of scientific evidence. But risk communication also includes expressions of opinion, concern, frustration, and the like by all participants, directed at whomever will hear and might act. Such decision making tends to be adver- sarial, with political actors making the strongest possible case for their positions, overtly expressing their interests and values or cit- ing expert judgment and analysis depending on which arguments seem most effective. Recipients of risk messages understand that those messages are guided by interests and political positions and so do not expect any single source to offer an unbiased assessment of available scientific knowledge. Public policy about tobacco smoking illustrates the range of risk messages that come out of public debate. The policy options for risk management involve decisions to be made in different bodies, each using different rules of debate and assigning different roles to the general public within those rules. For instance, the federal gov- ernment has considered increasing excise taxes on cigarettes, placing warning labels on cigarette packages, funding antismoking advertis- ing campaigns, distributing informational pamphlets on the health hazards of smoking, and banning smoking in various public places. Other options that might be considered for cigarettes, and that have been used for other health hazards, include outright prohibition on manufacture or sale and restriction to use by prescription only. In state and local governments, debates have also proceeded on op- tions such as banning cigarette advertisements in some public places, raising the minimum age for purchasing tobacco products, banning smoking in municipal buildings, and requiring no-smoking sections in restaurants. Risk communication varies from one of these decision-making arenas to another. Citizens participate in legislative settings by at- tempting to influence their representatives directly or by affecting the general climate of opinion and thus achieving indirect influence. In federal regulatory decision making, there is also wide latitude for participation, although the Administrative Procedures Act and agencies' practices constrain the time and type of participation and the kinds of arguments that can be introduced (Greenwood, 1984~. Agency procedures differ, particularly in terms of how much two-way
74 IMPROVING RISK COMMUNICATION communication they allow and how much they do to provide expert knowledge to the citizenry at large. Nevertheless, public debate in the regulatory or legislative context allows for risk messages and other related messages from a large number of sources. We consider risk communication in a setting of public debate successful to the extent that it raises the [eve! of understanding of rel- evant issues or actions among the affected and interested parties and those involved are satisfied that they are adequately informed within the limits of available knowledge. As noted in Chapter 1, successful risk communication does not imply optimal risk decisions; it only en- sures that the decisions are informed by the best available knowledge. Also as noted in Chapter 1, raising the level of understanding requires more than making accurate information accessible to the interested parties. Success requires increased understanding of the issues to the extent that the parties involved desire to understand. Although individual risk messages may contribute to increased understanding, the net effect of risk communication on understanding depends on all the messages individuals receive and their interpretation of them. Therefore, the designers of risk messages who wish to increase the recipients' understanding need to take into account the recipients' willingness and ability to receive and understand the messages as well as the effects of other, sometimes conflicting, messages that they may also receive. Success for risk communication does not require that every citi- zen be informed about the risks presented in every regulatory deci- sion, but people need to be confident that some person or group that shares their interests and values is well informed and is representing those positions competently in the political system. Public debate, in a traditional view in the United States, implies a pluralism of constituencies, with "consent of the governed" consisting of trust that the relevant views are represented, that the procedures do not disadvantage important constituencies, and that the people are able to hold public officials accountable for their actions. The requirement that interested parties believe they are ad- equately informed is worth explanation. It stems from recognition that in several arenas of public debate risk decisions are intensely con- troversial and many message sources are widely mistrusted. This sit- uation imposes requirements, particularly on those message sources and in those policy arenas, that may seem unfair to officials who
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 75 believe their responsibility to the public extends only to making wise decisions and providing complete, accurate information. But if a message source is widely mistrusted, its messages will be rejected by many regardless of completeness or accuracy. If accurate information is rejected by recipients, it does nothing to increase their knowledge base hence the requirement that recipients of information for public debate be satisfied that they are adequately informed. Both of the above-mentioned requirements for successful risk communication were factors in the public debate that resulted in the successful siting of the ECOFLO hazardous waste facility in Greens- boro, North Carolina. This siting case also illustrates an instance in which understandable and sensitive messages from an individual risk communicator (ECOFLO) contributed to the success of the overall risk communication process involving the Guilford County Hazardous Waste Task Force, environmentalists, and other concerned citizens (see accompanying story, pages 76-77~. Nevertheless, it should be emphasized that open and free communication will not necessarily ease conflicts in all situations. With respect to a designated decision maker, such as the head of a regulatory agency, risk communication is successful only if it ade- quately informs the decision maker. A decision maker is adequately informed within the limits of available knowledge if provision of all remaining available information would add nothing to justify a mod- ification of his or her choice. Decision makers need] to be informed about the managerial and political aspects of the choice at hand as well as about the state of technical knowledge. And, as already noted, the relevant knowledge should be understood by the decision maker, not merely made accessible. It is important to emphasize that a successful risk communication process is different from a risk message that is successful from the standpoint of its source. In a public debate (like that in the ECOFLO case), participants produce risk messages aimed at changing minds and influencing political outcomes. From their perspective a risk message is successful to the extent that it contributes to the outcomes its sponsor desires. Sometimes a risk communicator will make false or deceptive statements or will withhold pertinent information to achieve a political effect. Such activities, if they are not revealed, may achieve the ends of the message source but not the social goal of an adequately informed debate.
76 IMPROVING RISK COMMUNICATION ECO1?IO HAZARDOUS WASTE FACILITY SITING GREENSBORO, NORTH CAROLINA The successful siting of the ECOFLO hazardous waste facility in Greensboro, North Carolina, in 1985 is an example of good risk communication and effective risk messages. Although represent- ing a situation somewhat less problematic than those encountered elsewhere the company proposed a treatment facility to reduce the overall amount of toxic material in that locale it does illus- trate the role of communication efforts in the siting of a hazardous waste facility. The siting of such plants is notoriously difficult. As a result of ECOFLO's efforts, however, the final public hearing to site the facility lasted only 15 rn~nutes and led to the permitting of the plant with the blessing of local government officials and environmentalists (Lynn, 1987~. ECOFLO began operation in Greensboro in September 1983 with a license as a waste transporter. It worked mainly with small companies that produced about 20 drums of waste a month. Although ECOFLO was a new company, its owners had previ- ously worked for other hazardous waste companies. In July 1984, ECOFLO submitted its plans for a hazardous waste treatment facility to the state of North Carolina. The plant was designed to~serve primarily local and intrastate markets and would not handle PCBs, dioxins, cyanide, radioactives, biological wastes, or explosives. The treatment processes to be used were neutraTiza- tion and centrifugation. Wastes that had to be burned would be transported elsewhere (Lynn, 1987~. A year and a half prior to ECOFLO's application, another company had tried to site a hazardous waste facility in Greensboro and failed. Local citizens, unable to receive information or to have their concerns addressed, had successfully organized opposition to that facility. The Greensboro area had a group of citizens well versed in hazardous waste issues. As a result of an EPA grant to the North Carolina League of Women Voters in 1979, the Guilford County Hazardous Waste Task Force was formed. The task force spon- sored short courses on toxic materials and workshops and displays to educate and organize the community. By 1985 the task force and its chair, Carolyn Allen, had good working relationships with the local government staff and elected officials.
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 77 When ECOFLO decided to site in Greensboro, the task force invited neighborhood leaders from the part of the city where the facility might be located to a series of education meetings on hazardous waste. These workshops included the chemistry of haz- ardous waste, disposal processes, and a session with Tom Bar- bee, ECOFLO's vice president and the Greensboro plant manager (Lynn, 1987~. This was not Barbee's first contact with the task force. He had been attending task force meetings since 1979, as a profes- sional waste manager with another firm. He was also a native of North Carolina and a longtime Greensboro resident. He did not see the environmentalists as the enemy. In a local TV interview he said that ECOFLO "honestly wants to be a service to the community.... We want to help local companies handle their waste as responsibly as possible.... We are on the side of the environmentalists" (quoted in Lynn, 1987~. From the time ECOFLO decided to site a facility in Greens- boro, Barbee had been contacting relevant groups and individuals. He went to the local police and fire departments to ask what they thought he needed to do to ensure a safe site. He talked with ministers, neighbors, the planning and zoning department, and county commissioners. He gave candid and detailed answers to questions by citizens. He and his staff took the press, state and local officials, and neighbors on plant tours. He even sponsored his own public meeting before the state held its public hearing. Barbee's meeting was cohosted by Bruce Banks, a local chemistry professor and Audubon Society member; Carolyn Allen, chair of the task force; ant! Jim Rayburn, chair of the Guilford County Advisory Board on Environmental Affairs (Lynn, 19873. At this meeting Barbee detailed how he had made changes in his original proposal based on feedback from the fire department, the planning commission, and the task force, among others. He invited public participation and took the public's concerns and suggestions into consideration in ECOFLO's revised plan. This willingness on the part of ECOFLO to involve the com- munity, to share information, and to implement changes based on community input proved effective. The ECOFLO waste treat- ment facility was approved and the citizens were satisfied it could be operated safely (Lynn, 1987~.
78 IMPROVING RISK COMMUNICATION Personal Action Risk communication regarding personal action is quite different from risk communication regarding public decisions. At minimum the setting is more limited because most risk messages are addressed to individuals rather than to a spectrum of participants in public de- bate. Sending messages to an individual is in one respect like sending them to the head of a regulatory agency: both have the ultimate au- thority to act. But the two situations are also different in important respects: few individuals have staffs of experts paid to answer their questions, and individuals seldom want the amount of detail that is justified when a federal regulator is about to make a decision for the whole population (see Figure 4.1~. Much of risk communication in this setting takes the form of messages directed at the public offering information, advice, warnings, or recommendations regarding risky individual actions. Both public agencies and private organizations sometimes design such risk messages. But personal action is also influenced by a variety of risk messages, usually informal, from other individuals. People want to know how hard it was to stop smoking, or whether low-fat meals can be made to taste good, or in what ways other people feel better after losing weight. Such risk-related messages, regardless of whether they accurately represent the likely outcomes of alternative actions, can be critical in individual decisions (Nisbett and Ross, 1980~. Tobacco smoking also illustrates the kinds of risk communica- tion issues that arise in the context of personal choice. Despite the restrictions created by recent policies, people still choose whether, how much, when, and where, within limits, to smoke. But Congress has decided that it is in the public interest to influence smoking behavior in various ways short of directly restricting tobacco use. Cigarette taxes and advertising restrictions are two policies that constrain individuals and the tobacco tracle. Other policies, such as the requirement of warning labels and widespread dissemination of the surgeon general's findings on the risks of smoking, rely on risk messages as an alternative to direct control of the substance. Such policies create a risk communication setting much different from that of public decision making, particularly because they call for special- ized risk messages. Congress has sanctioned efforts by government officials, including the surgeon general and other medical experts, to design and disseminate messages aimed at changing individual behavior. We consider risk communication in the setting of personal choice
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 79 A /, 4-~: ~ I\ / 11 'if it, < 5 ~ 1n~6 >I NEXT \ . _ , 1 ~ FIGURE 4.1 For personal action to reduce risks, a simple warning sign (e.g., Chills and Curves Next 10 Vilest may be sufficient; a report of a formal risk analysis could be counterproductive. SOURCE: Courtesy of Paul Stern. successful only If it adequately informs the individual for making a choice among alternatives. Adequate information, to reiterate, must be understandable for risk communication to succeed; it is not sufficient that it be available. Part of the debate is about going further, so that the recipients are somehow brought to understand the material. But we have not gone so far as to include this an a criterion for success. Getting recipients' attention and comprehension poses signifi- cant barriers to risk communication, especially in the arena of per- sonal action, where many recipients customarily act without care- fully considering risks and benefits. It should be noted that from the standpoint of the designers of risk messages, the goal may or may not be to inform choice. Often a message is intended to influence choice, a very different matter, even if experts believe that the choice they desire to elicit is in the audience member's interest. Thus some risk
80 IMPROVING RISK COMMUNICATION messages from government agencies are designed to inform choice (e.g., nutritional information on food packages), but at other times, occasionally after open debate in a legislative setting, an explicit decision is made to influence beliefs or behavior in a particular direc- tion (e.g., anti-drunk-driving campaigns). Although risk messages are sometimes judged against a criterion of behavior change, this is not an appropriate test of whether an individual has made an in- formec! choice. It is possible for an individual, fully informed of the risks, to choose to engage in hazardous behaviors such as smoking, skydiving, or leaving seat belts unbuckled. Sometimes risk messages are intended to inform or explain rather than to be used as direct input to a choice. This can be the case when the risk manager is in the position of explaining a decision that has already been made. It can also occur in situations when individuals or groups are unavoidably exposed to particular hazards. It may be necessary to explain why a decision has been made that is injuri- ous to the recipients of the message or that has other undesirable consequences. INFORMATION AND INFLUENCE: THE PURPOSES OF RISE MESSAGES We have noted that successful risk communication, such as that described in the ECOFLO case, makes for better-informed decision makers, both individuals and public or private officials. A "success- ful" risk message, in contrast, is not always one that increases the understanding of decision makers. For risk messages success is com- monly interpreted in relation to the goals or purposes of the message source. The sources of risk messages sometimes aim to inform the re- cipients, but sometimes they aim to influence their beliefs or actions. A risk message designed to influence may be judged successful even if it does nothing to add to the audience's understanding. An antidrug campaign that relies on exhortations from prominent sports figures is successful if it keeps some teenagers from addiction, even if they learn nothing new about the health effects of heroin or cocaine. We recognize that efforts to influence through risk messages do not always have such noble purposes. The sources of risk messages may set their own criteria of success but attaining them does not always advance a public good. Sometimes "effective" risk messages are inconsistent with promoting substantive public good, as when they mislead people about what is in their interest. At such times
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 81 they are in conflict with the public goal of successful risk communi- cation. (Sometimes, however, audience members gain understanding even from biased risk messages. For instance, judges, elected officials, and interested citizens often gain understanding on matters of public controversy by comparing messages from various sources that they realize are trying to influence them. They inform themselves, despite the efforts of message sources to influence rather than inform.) Serious confusion can arise because any given risk message may be intended to inform or to influence. It can be difficult for a recipient to tell which aim a particular message has; message sources, aware of this difficulty, sometimes attempt to persuade in the guise of informing. That tactic is likely to be most effective when it goes undetected, but it can backfire seriously if revealed, undermining the credibility of the message source and creating resentment and mistrust. The problem of dual purposes is compounded by the fact that the designers of risk messages are often called on to both inform and influence the same audience with the same message. Regulatory agency employees, for instance, are routinely asked to prepare a document to support a decision at the end of a formal ruTe-making process that both summarizes the evidence on which the decision was based (thus informing the audience) and justifies that decision (thus endeavoring to influence the audience to believe the right choice has been made). The dual purposes of risk messages complicate defining respon- sible behavior for the designers of the messages. In order to arrive at some criteria for the acceptability of attempts to influence, we begin by describing a dimension along which one can array techniques for the construction of risk messages. At one end of the dimension is an ideal, pure information, free of techniques of influence; at the other end is deception. Although the purpose of informing is consistent with the goal of successful risk communication to raise decision makers' level of understanding-the use of techniques that aim to persuade, deceive, or otherwise influence decision makers implies that a different goal is being pursued. l~fo~mation To inform someone about an issue or choice is to assist that person to apprehend the relevant propositions or statements that describe the issue or choice. Ideally, the result is that the person or persons informed gain a full or complete understanding of the issue
82 IMPROVING RISK COMMUNICATION or choice. This appears to have happened in the ECOFLO case. In practice, however, full understanding does not exist for most impor- tant choices about risk (see Chapter 2), so it cannot be conveyed. A practical goal for information is for the recipient to gain understand- ing, within the limits of available knowledge' that is adequate to make appropriate choices given his OT her values. Adequate understanding does not require knowing everything that is known about an issue, only enough to be able to make choices in one's own best interest. If more precise information would enable members of the audience to make choices that better approximate their desires, it should be provided; if it would not aid in decision making, more precision is unnecessary. Influence A spectrum of techniques is available for designing risk messages that go beyond pure information and that can be used to influence an audience. The most extreme techniques involve outright deception: strategies such as "lying, withholding of information, true assertion that omits a vital qualification, and misleading exaggeration to cause persons to believe what is false" (Faden and Beauchamp, 1986:363~. But many influence techniques do not do such violence to the truth. In order to consider the appropriateness of different techniques, it is useful to identify them. The following paragraphs describe different techniques, beginning with some that stay close to the facts and moving to some that do not depend much on factual information. Some of these techniques can be used either to inform or to influence. It is this possibility that makes it difficult for recipients of risk messages to determine their intent and therefore to interpret their content. Highlighting Facts Risk messages cannot include all the details known to science and still be read and understood by most nonexperts. Therefore the designers of messages omit some information and highlight other in- formation. For instance, message designers choose whether to sum- marize knowledge about both possible deaths and illnesses arising from a risk or only about deaths, about both direct and synergistic ef- fects or only direct effects, about effects on subpopulations including sensitive groups or just on whole populations, and so forth. Having
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 83 chosen what to present, message designers must also make choices about what parts of the message to emphasize with visual aids, vocal emphasis, underlining, color, and other techniques. Although high- lighting may be employed only to emphasize the essentials of what is known, decisions to highlight which are unavoidable- involve judgments about what is essential. A large psychological litera- ture demonstrates that highlighting information, or making it more "available," affects the understanding and the decisions of those who receive the messages (Fiske and Taylor, 1984; Kahneman et al., 1982; Tversky and Kahneman, 19733. Thus highlighting can influence the audience's beliefs about what aspects of a risk decision are important in the direction desired by the message designer. Framings Information and Decisions Different ways of presenting the same facts can create different impressions. When a risk estimate is uncertain, it can be described by a point or "maximum likelihood" estimate or by a range of pos- sibilities around the point estimate. But estimates that include a wide range of uncertainties can imply that a disastrous consequence is "possible," even when expert opinion is unanimous that the like- lihood of disaster is extremely small. The amount of uncertainty to present is a judgment that can potentially influence a recipient's judgment. Another example of "framing" involves the choice between al- ternative ways of presenting the same numerical information. One study, for example, found that a hypothetical vaccine that reduces the probability of contracting a disease from 0.20 to 0.10 is less attractive if it is described as elective in half the cases than if it is presented as fully effective against one of two virus strains that strike with equal probability and that produce the same disease. This finding suggests that people favor full protection against an identified risk over equivalent but probabilistic protection (Tversky and Kahneman, 1981~. Sirn~lar differences in presentation have been identified with respect to whether outcomes are presented in terms of "sure Toss" or an "insurance premium" (Fischhoff et al., 1980) or "lives lost" as opposed to "lives saved" (Tversky and Kahneman, 1981~. It has even been demonstrated that when two versions are presented sequentially people often reverse their preference from the first presentation to the second (Hershey and Shoemaker, 19803.
84 Risk Comparisons An important instance of framing is the use of risk comparisons. Comparing one risk that is not well understood to another that the audience comprehends may be a useful way to convey information about the former risk. It is often difficult, however, to find risks that are similar on enough attributes to carry the comparison. But risk comparisons can also be used to influence or even misread, because a risk comparison may improperly carry the implication that if a person is willing to take the larger of two risks he or she should accept the smaller as well (Covello et al., 1988; Fischhoff et al., 1981a). The uses of risk comparisons are discussed in more detail in Chapter 5. IMPROVING RISK COMMUNICATION Persuasive Use of Facts Risk messages often involve a selection of the facts to make a point. Messages aimed at convincing recipients of a point of view can use techniques of highlighting and framing but can also employ other rhetorical techniques: selective presentation of evidence, creation and destruction of "straw-man" arguments, judicious placement of the various arguments within a message for maximum effect, listing of supporting arguments by number to make the argument look stronger, and so forth. Such techniques can enhance the persuasive effect of messages, sometimes without any alteration of the content (Cialdini, 1984; Eagly and Chaiken, 1985; McGuire, 1985), and they can be quite difficult for a recipient to detect. Appeals to Authority Nonexperts often want to know who has taken what position on a difficult choice before them. When they do not know enough to make an informed choice themselves, or believe it too expensive or time consuming to become fully informed, they may choose to adopt the position of a person or organization they consider expert and trustworthy. Thus risk messages can be influential by supplying information about who has taken positions on an issue. They may be balanced in their references to authority or they may not: a message may quote some scientists in support of a position but omit quotations from similar scientists who disagree. They may quote
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 85 relevant authorities who have specialized knowledge or they may refer to sources widely trusted on other issues but ill informer] on the issue at hand. And they may be accurate or inaccurate in representing the views of the authorities. Clearly, appeals to authority can fall at many different points along the dimension from pure information to deception. A ppeals to Emotion Risk messages sometimes appeal to fear, pride, guilt, commu- nity spirit, parental concerns, or other emotions to spur people to action. Sometimes emotional appeals are made in the context of a presentation of information. Thus, saying that cigarette smoking causes emphysema conveys the same information with or without an accompanying film of an end-stage emphysema patient, but with the film the message will have a different effect. Appeals to emo- tion are not always more effective in inducing behavior change than less emotional appeals: the psychological research shows that the effect depends on other aspects of the message as well (Petty et al., 1988~. Nevertheless, appeals to emotion can be effective influence techniques under some conditions. Sometimes the use of emotional appeals is widely accepted, but often it is considered manipulative and irresponsible. The conditions under which emotional appeals are considered acceptable are not well understood. ~. ~. ~ USE OF INFLUENCE TECHNIQUES IN RISE COMMUNICATION Achieving Balance Risk messages often employ some of the above influence tech- niques; indeed, it is difficult to imagine a risk message that could attract the attention of nonexperts without making use of at least highlighting or framing. A paradox arises for risk communication: How can messages be made to improve the recipients' base of in- formation if, in order to be effective, they must use techniques of influence? The paradox disappears when one realizes that there are strategies for controlling the use of influence techniques consistent with the goal of successful risk communication. Substantive guide- lines should be established for the content of risk messages that responsible message designers, including government officials, can
86 IMPROVING RISK COMMUNICATION to keep influence techniques under control so as not to bias recipi- ents' understanding. Because available knowledge is inadequate to provide highly detailed substantive guidelines, procedural approaches that keep message designers in bounds are also critical to achieving successful risk communication. The strategy of substantive guidelines is highly demanding. As already noted, the language of risk messages and even the measures used in risk analysis often embody value judgments or otherwise tend to lead the recipients of messages toward particular conclusions. We have noted several examples, but not enough is known to identify all the ways a risk message might bias a recipient's understanding. Thus it is not now possible to devise a complete guide to sources of potential bias that would allow risk messages to be evaluated for bal- ance. Moreover, research on communication strongly suggests that the most effective message design for any particular purpose varies with the subject matter at hand, the decision alternatives, the in- tended audience, and other factors. But very little is known about the key situational variables that alter the effects of risk messages. Thus at present any guidelines for balanced risk messages would lack situational specificity. Existing knowledge can help message design- ers by identifying some potential pitfalls, but it cannot yield highly specific guidance. Responsible message designers need to interpret available advice, keeping in mind that knowledge is incomplete and that general principles may not apply to certain specific situations. Since there is no clear best way to make such judgments, substantive guidelines are not enough to ensure balance in risk messages, even when the sources are doing their best to achieve it. The procedural strategy, which relies on a system of checks and balances to control the possible biases in risk messages, is applicable without regard to the state of knowledge about the effects of risk messages. The strategy assumes that available guidelines will never be perfectly correct or clear-cut and that vested interests or strongly held values will often induce ingenious message designers to find ways around guidelines. It therefore relies on systems of scrutiny and criticism, and the discipline of competing messages, to keep message designers within bounds. Examples of procedural strategies applied to individual messages are the procedures of the National Center for Toxicological Research (NCTR) Consensus Workshops and those of the National Research Council (NRC) for review of its reports. The NCTR Consensus Workshop Series involves scientists from academia, government, in
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 87 dustry, and public interest groups gathered to resolve toxicological issues, usually concerning the hazard posed by particular substances (Gough et al., 1984~. Consensus is sought, not by formal voting, but through the chairman's guiding discussion toward agreement. Careful procedures ensure that all panelists have an opportunity to submit statements and to evaluate and comment on reports. These procedures ensure that reports focus on those areas where consensus is reached and present the major factors in reaching agreement. The NRC, many of whose reports are detailed messages about risk, does not rely on guidelines for the use of language, graphics, and so forth. Rather it relies on a balanced choice of committee members and an independent review process. The NRC presumes that a dialogue of well-informed individuals with varying perspectives will yield a first approximation of a balanced assessment. The outcome of this pro- cess is double-checked by submitting it to an independent review process involving experts who also represent a range of perspectives. In these two procedures it is not substantive guidelines but the process of dialogue and criticism that is used to ensure a balanced message. Achieving Influence Even more difficult than the problem of achieving balance in risk communication is the problem of deciding whether balance is the wrong objective. Advocates whose clear purpose is to influence their audiences may experience no problem, but the issue can be particularly acute for public officials who sit in a relation of public trust to the recipients of their messages. When should messages aim at merely informing the public, or government decision makers, and when should the goal be to influence the recipients? Government officials are commonly expected to follow a more re- stricted standard of behavior in the area of risk communication than are advocacy groups, private citizens, or corporations. Similarly, cit- izens apply a stricter standard to messages paid for with public funds than to privately funded messages. We judge that such standards are justified because government officials hold a public trust. But the specifics of such standards are not easily defined. After considerable debate focusing on the appropriate use of risk messages by public officials, we concluded that no explicit guidelines can be drawn defining which techniques are appropriate or inappro- priate in particular situations or for particular message sources. We
88 IMPROVING RISK COMMUNICATION agreed that informing is always an appropriate goal in the design of risk messages and that deception is never appropriate. But we recognize that messages that employ influence techniques or that have influence as an objective are often considered acceptable, even coming from public officials. We believe that more extensive public debate is needed to arrive at standards for responsible behavior by public officials in the design of risk messages. As a contribution to that debate, we offer the following observations about the conditions under which influence techniques seem most likely to be considered appropriate by various audiences. First, the acceptability of influence as a purpose of risk messages seems to depend in part on which beliefs or actions are being influ- enced. Consider the range of actions and opinions that government agencies have tried or might try to influence with risk messages. Here are some examples: . swine influenza; Inoculating children against diphtheria, polio, pertussis, or Using condoms to prevent AIDS, gonorrhea, or pregnancy; Avoiding or reducing consumption of heroin, alcohol by drivers, tobacco products, alcohol by pregnant women, aspirin by children, or animal fat; Using seat belts, motorcycle helmets, or masks for painting or working with fiberglass; . Supporting drug enforcement activities, AIDS research, EPA enforcement activities, or the repeal (or passage) of particular pieces of legislation. . Depending on the action or opinion in question, the likely re- sponse to government-sponsored influence attempts may vary from general acceptance to extreme controversy. Within each of the cat- egories just listed, we believe that efforts to influence the action or opinion mentioned first would be relatively uncontroversial compared with similar efforts to influence the actions given later in each cate- gory. It is important to recognize, however, that observers, including members of our study committee, differ on the appropriateness of influence techniques in certain of the contexts listed. Some variation in judgments concerns scientific knowledge: the more clearly it has been established that an activity is dangerous or that it may harm persons generally considered to deserve societal protection (e.g., chil- dren), the more acceptable influence attempts seem to become.2 But
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 89 because of scientific uncertainty, informed observers sometimes dis- agree about how well established the relevant knowledge is. Another central issue seems to be the compatibility of governmental influence with individual autonomy and related values (Fader, 1987; Faden and Beauchamp, 1986~. When a class of personal action (such as drunk driving) affects a large portion of the populace or threatens to inflict substantial monetary and other costs on society or on in- dividuals who do not engage in that action, people are more willing to accept, and even to demand, that government agencies be proac- tive and try to influence beliefs and actions. Under such conditions, people are more willing to compromise the autonomy, privacy, or freedom of some individuals for the good of others. Second, the acceptability of influence seems to depend on the techniques employed. Generally, the farther an influence technique lies along the dimension from information to deception, the harder the message becomes to justify and the clearer and more explicit must be the legitimate public purpose being served. To influence people to use condoms to prevent AIDS, government might appeal to authorities (the surgeon general recommending use of condoms to avoid AIDS) or respected or admired individuals (fiIm and popu- lar music stars hosting a TV special encouraging use of condoms in AIDS prevention), post warning signs (in lavatories of establishments frequented by homosexual males), present selected risk and risk re- duction information ("use of condoms can reduce the transmission of AIDS by 95 percent"), or appeal to emotion (photographically depict the late stages of AIDS or state that "you sleep with your partner's whole sexual history"~. Observers differ on the appropriateness of such techniques for a particular purpose, even when all agree that the purpose justifies some form of governmental influence. We conclude that public values about the importance of par- ticular public purposes and the acceptability of particular influence techniques are not well understood. Generally, the more an influ- ence attempt would compromise important values such as personal autonomy or constitutional guarantees such as freedom of speech or association, and the more closely the influence technique approaches deception, the more it needs to be legitimated in order to be accept- able. Legitimacy is what makes people consider a particular influence attempt either responsible or irresponsible and either appropriate or inappropriate for government officials. But there are no clear a priori guidelines that can tell a govern- ment official or other designer of a risk message when the message's ~1 . ~·.
90 IMPROVING RISK COMMUNICATION purposes are sufficiently legitimate to justify a particular technique that goes beyond informing. Government officials will likely find their efforts to influence contested if they stray from accepted scien- tific views or if they challenge popular consensus. It is for this reason that decisions about governmental use of influence techniques in risk messages are often debated in overtly political arenas rather than being left to unelected officials' unscrutinized discretion. We believe that political arenas are the proper place for deciding the appropri- ateness of governmental efforts to influence citizens. Governmental attempts to influence citizens' beliefs and actions can be justified only to the extent that some legitimate public process has culminated in a decision that using risk messages to influence behavior serves an important public purpose. Influence and Personal Action The clearest example of politically established legitimacy for risk messages occurred in the congressional debate on persuading people to stop smoking. A congressional act codified language-a set of risk messages that now appears on cigarette packages. The process of debate and approval by elected officials granted legitimacy to the messages. Such explicit public (rebate rarely occurs to give clear prior jus- tification for governmental attempts to influence personal behavior. Nevertheless, an agency or official can sometimes act legitimately on general authority. For example, public health officials have fairly general support in the mandates of their agencies for influencing people to take action to prevent the spread of infectious diseases. As a result, the surgeon general's 1988 mass mailing of a risk mes- sage about avoiding AIDS was met with wide public acceptance and even gratitude. Sometimes executive branch officials justify Influence attempts within the spirit of their legislative mandates. The U.S. En- vironmental Protection Agency's efforts to inform the public about the health risks of indoor radon, and to convince people to have their homes tested and sometimes modified at considerable expense, are not justified by anything stronger than the EPA's general mandate for environmental protection. Yet this attempt to influence behavior in the setting of personal action was widely welcomed.
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 91 InJ7uence and Public Debate Sometimes executive branch officials rely on their general man- date to influence beliefs in the setting of public debate. Such efforts tend to be more acceptable after a risk management decision than before (e.g., when regulators are expected to justify their decisions to the public). But even before a decision is made, there are situ- ations in which some kinds of efforts to influence public debate are appropriate. Regulatory officials sometimes argue that they have an obligation to evaluate new risks and, when public action is needed, to persuade elected officials of that fact. It is not enough, they say, merely to inform the public of the latest knowledge. Thus some pub- lic officials, on receiving evidence on the risk to the earth's ozone layer from chiorofluorocarbons, attempted to influence the highest levels of government to support an international treaty to cut production of that class of chemicals. But it is easy for a public official to overstep the bounds of acceptability. This happens most readily when the subject matter of the influence attempt is already politically controversial or when government can be seen as trying to influence free political expres- sion. When the San Francisco office of the Energy Research and Development Administration distributed 78,000 pamphlets defend- ing the safety of the nuclear power industry during a 1976 California referendum campaign on the future of nuclear power, the result was a critical report from the General Accounting Office and strong ex- pressions of congressional outrage (Burnham, 1976~. Not only was the message unacceptable, but its dissemination and the agency's evasive response to criticism harmed the agency's credibility. With many influence attempts it takes fairly explicit debate and agreement to make them legitimate: vague appeals to an agency's mandate are not sufficient. The judgment of whether public officials have or have not ex- ceeded their proper role in a particular attempt to influence public debate is difficult to make. But it is a matter of judgment. Clearly, the freedom of public servants to influence decision makers must be kept within bounds. We considered and rejected the position that advocacy is always inappropriate for executive branch officials in the setting of public debate. There are situations in which such officials are in the best position to alert the public to a hazard that may deserve governmental action. But it is difficult to define the proper
92 IMPROVING RISK COMMUNICATION limit. Scientific analysis is indispensable to successful risk commu- nication. It~can show what is known about risks and the attendant choices and can identify the limits and uncertainties of that knowl- edge; it can therefore indicate what can be said. Science can also advise on when and how best to say it in order to improve an audi- ence's understanding or to influence beliefs and actions. A decision to engage in advocacy, however, involves judgments about which risk management option is appropriate and about how much to influ- ence audiences with other than information judgments that must be based on values as well an knowledge. We concluded that natural and social sciences cannot provide guidelines for when to engage in advocacy in risk communication. Although empirical research can determine which beliefs Americans consider acceptable for influence by government and which influence techniques they consider most extreme and therefore most in need of legitimation, there is no prac- tical way to tell in advance whether enough legitimation exists in the political system to justify a particular attempt to use risk messages to influence recipients. Advocacy messages from executive branch officials must therefore be judged against the legitimate role of the officials in question, as set forth in the relevant legislation and judi- cial interpretations and as argued by elected officials. The decision of what are legitimate bounds for governmental risk messages is and ought to be made through the political process. We recognize that the boundaries for advocacy in the political process often are clear only after a public official has overstepped them, leaving public officials in an unpleasant position. However, such boundaries usually can be discerned in advance by careful anal- ysis. In any case, when officials judge that the public welfare depends on a specific change in policy or individual behavior, they must also judge how far they can go before overstepping legitimate constraints. Advocacy can be politically risky for public officials. It may be widely applauded or widely condemned, and types of messages that may be widely accepted on one subject matter or from one govern- ment source may be criticized when the topic or source changes. A public official should be aware of the political risks and of the legiti- mate constraints placed upon government in advocacy and, where an unusually strong degree of advocacy seems warranted, seek political approval of such action. Risk communication may be difficult because the purposes of messages are not clear or because they have multiple, perhaps con- flicting, purposes. The next chapter describes several misconceptions
PURPOSES OF RISK COMMUNICATION AND RISK MESSAGES 93 about risk communication that may also contribute to confusion and frustration on the part of risk communicators and recipients. NOTES 1. Generally, persuasive messages are less effective when recipients have the opportunity to "anchor" their preexisting beliefs against persuasion in the following ways: by defending them against a prior persuasive message, by considering their other beliefs or values that are supported by the belief subject to persuasive communication, or by training in the ability to question or argue against persuasive messages or to be suspicious of the source (the evidence is reviewed by McGuire, 1985:292-294~. Persuasion that does not appear to be persuasion might not evoke such defenses. 2. For instance, public support for persuasive messages about AIDS pre- vention was minimal when the disease seemed to threaten only homosexual males, Haitians, and intravenous drug users but increased rapidly when chil- dren, hemophiliacs, adult heterosexuals, and hospital patients receiving blood transfusions were seen to be at risk. Shilts (1987) gives an extensive account of how public concern about AIDS has related to the identity of the groups believed to be" at risk.