Tobacco Documents Online

x Comments by W. Kip Viscusi on FDA Notice of Findings, "Regulations • Restricting the Sale and Distribution of Cigarettes and Smokeless Tobacco Products to Protect Children and Adolescents: Findings of the Focus Group Testing of Brief Statements for Cigarette Advertisements," 60 Fed. Reg. 61,670-79 (Dec. 1, 1995). ~George G. Allen Professor of Economics, Department of Economics, Duke University, Durham, NC 27708-0097, phone 919-660-1833, fax 919-684-8974.

• • Backaround I have been asked by Covington & Burling, which represents the Tobacco Institute, to prepare comments on the FDA notice of December 1, 1995, pertaining to the sale and distribution of-cigarettes and smokeless tobacco. I am currently the George G: Allen Professor of Economics at Duke University and have published 15 books and over 150 articles, many of which deal with the issues addressed in this rulemaking. For example, many of these books focus on risk perceptions and product safety. One of my recent books and several recent articles focus on cigarette smoking. My books have won three successive book of the year awards of the American Risk and Insurance Association. I am also the founding Editor of the Journal of Risk and Uncertainty and currently serve on seven other editorial boards, including the American Economic Review. An additional summary of my background and my C.V. and bibliography are appended to this statement. Overview The FDA notice of December 1, 1995, arises out of the Agency's proposal to require an additional warning message on cigarette advertising to help discourage smoking by persons under the age of 18. In my comments, I address three ways in which FDA's proposal is seriously flawed. First, the focus group fmdings are incapable of demonstrating that the proposed warnings would advance the Agency's goal, because focus group studies are merely preliminary steps to scientifically valid testing of the effects of a hazard warning and cannot by themselves serve as a test of any hypothesis about these effects. Second, independently of the methodological defects in FDA's focus group approach, there is no reason to believe that any additional warnings about the risks of smoking could achieve FDA's goal of reducing youth smoking by 50 percent, because youths already overestimate the health risks of smoking. Third, FDA's warnings that the -2-.

s • s • teenagers in the focus groups seemed to favor may have unintended adverse consequences -- on youth smoking rates in particular, and on the development of a consistent hazard warning vocabulary in general. The Focus Group Approach The foundation of the FDA notice is based almost entirely on results of focus groups undertaken with samples of 12- to 17-year-olds by a survey research firm, Macro International. l Focus groups are, of course, a frequent part of survey design. However, the purpose of focus groups is not to test hypotheses, but rather to serve an exploratory role in, for example, developing question language for a survey. It is typically a preliminary stage of the survey design that takes place before pretesting of a survey instrument and before undertaking a survey that would have any scientific validity. Focus group reports are not a substitute for a scientific study or a controlled experiment that analyzes hazard warning labels; the entire character of the focus groups is simply not designed to make it possible to undertake scientific hypothesis tests. Indeed, the report that Macro International prepared for FDA repeatedly emphasizes the limitations inherent in focus group testing as a predictor of the effects of hazard warnings and explicitly cautions that the results are merely preliminary to further, scientifically valid studies: "This study cannot be considered reliable or valid in a statistical sense since the recruiting of participants cannot be replicated, nor can the moderator ask the same questions of other respondents. Certain biases are inherent in this type of study and are stated here to remind the reader that focus group data cannot be projected to any universe of individuals. 1 As noted in the background appendix, I have served as a consultant to Macro International on warnings issues. 3

9 • • • Finally, the reader is reminded that this report is intended primarily to clarify cloudy issues and point the direction for future research, and that data here cannot be projected to a universe of similar respondents. " 60 Fed. Reg. 61,672. Although these caveats stress the limitations of the population mix, the entire focus group process lacks solid foundation. The contrast between the approach that FDA undertook in this instance and the approach that I have taken on behalf of EPA reflects the differences between a focus group and a scientific study. In our studies of warnings for EPA, which were subsequently published as books by Harvard University Press and MIT Press,2 Wesley Magat and I also engaged in a pretesting phase. However, the results of this pretesting were then used in conjunction with evidence from the literature on warnings to design a series of.experimental treatments, one of which involved no risk information, and others of which explored different types of warning content and format in providing the warning message. Participants in our study answered the questionnaire giving detailed demographic information, making it possible to statistically control for individual differences. Moreover, this information served to establish a baseline with respect to respondents' risk beliefs and other aspects of their behavior. Respondents were then provided with the hazard warning that was pertinent for their particular experimental manipulation. The effect of the experimental treatment was assessed by investigating how risk perceptions and various measures of precautions are altered as a result of different hazard warnings. Because precautionary intentions may not always correspond with actual behavior, it is important to check the validity of the experimental design. We established this 2 W. Kip Viscusi and Wesley Magat. Learning About Risk: Consumer and Worker Responses to Hazard Information Cambridge: Harvard University Press, 1987 and Wesley Magat and W. Kip Viscusi, Informational Approaches to Regulations, Regulation of Economic Activity Series No. 19, Cambridge: MIT Press, 1992. 4

0 • 0 validity through a separate survey involving a different sample in which precautions were assessed for a product with a hazard warning label that directly corresponded to the risk information content and format of one of the labels used as an experimental treatment. The failure to find any significant differences in behavior between actual labels on the market and the experimental analog suggested that the study did a meaningful job of assessing how warnings would in fact influence behavior. Using these results, we estimated statistically the effect of hazard warnings on risk beliefs and various measures of precautionary behavior. Thus, it was possible to test scientifically whether in fact warnings improved risk judgments and whether they affected behavior. The purpose of hazard warnings is to promote sensible choices, and as a result it is essential to ascertain whether the information is being correctly understood and whether it produces accurate risk assessments and appropriate risk-averting decisions. In contrast, the FDA study went no further than the pretesting phase. FDA does not contemplate conducting the additional, research studies that FDA's own consultants have warned should follow the focus group studies. Instead, FDA states that it "will use the report, as well as comments submitted on the report, to determine the design, format, and content of the brief statement when preparing a fmal rule." 60 Fed. Reg. 61,670. FDA appears to have accepted the opinions of select groups of teenagers about the effects of the Agency's proposed warnings as valid data even though they have not been validated by scientifically rigorous testing. The design of hazard warnings is not an exercise that can be delegated to amateurs or, for that matter, to teenagers. To ask children aged 12 to 17 to, in effect, design a hazard warnings policy for cigarettes is completely irresponsible. FDA has ignored the considerable volume of academic literature that has developed guidelines for hazard ~ O warnings policy that could have assisted the Agency in evaluating the design of C~ ~ C ~ W -5- ~ GO

• • • meaningful and effective warnings. This literature goes beyond the many books and articles by myself and also includes a wide range of other studies from researchers in a variety of disciplines.3 One of the conclusions that FDA could have drawn from this literature is that, in any assessment of the efficacy of warnings based on responses by a sample of representative consumers, such as teenagers, it is essential to design the study and undertake it in a manner that utilizes their responses to test scientifically the efficacy of warnings rather than, in effect, delegating the responsibility of making policy decisions to children. FDA's approach represents a departure from the scientific literature in that it included no substantive testing of the efficacy of warnings. SmokinQ Risk Beliefs The underlying assumption that provides the impetus for FDA's proposal to require additional warnings is that youths are not aware of the risks of smoking or do not sufficiently appreciate the seriousness of those risks. If this assumption is incorrect, then the additional warnings that FDA has proposed will not achieve the Agency's goal of reducing underage smoking. Warnings can change behavior, but unlike regulations, warnings do not demand obedience under threat of sanctions but instead communicate information. It follows that warnings can only change behavior by providing relevant information of which an individual was previously unaware. Therefore, regulatory agencies should only use warnings in situations in which individuals are acting with inadequate information or in which there is substantial misinformation that the government is trying to correct. Warnings that instead attempt simply to browbeat individuals into changing their behavior will not be successful. O 3See not only the 1987 survey by the U.S. Department of Health & Human Services but ~ also the American Law Institute, Enterprise Responsibility for Personal Inj,ury-Reporters' ~ Study, Philadelphia: American Law Institute, 1991. © Qa c.~ -6- ~ co tc

• • Although underage smoking is a legitimate object of concern, inadequate risk perceptions are not its cause. Adolescents smoke not because they are unaware of, or underestimate, the risks; virtually all adolescents believe smoking is dangerous and start to smoke in spite of their beliefs about the hazards involved. Because the problem is not some lack of information or any defect of awareness, additional warnings can serve no purpose as educative tools and hence are unjustified. It is useful to review my published evidence on risk perceptions by those in younger age groups. Based on the most recent estimates of the lung cancer mortality risk to smokers from the data provided by the U.S. Surgeon General and the National Cancer Institute, I have estimated the lung cancer mortality risk to smokers as being .06-.13.4 This scientific evidence provides the reference point for assessing whether individuals' subjective risk perceptions are accurate. For all age groups, the overall assessed lung cancer risk is .43.5 Younger respondents do not assess the risk as being lower, as the FDA might lead us to believe; as it turns out, respondents aged 16-21 assess the lung cancer risk as being .49, and respondents aged 16-21 who smoke assess the lung cancer risk as being .45.6 These risk perceptions exceed those of adults and dwarf the estimated lung cancer risks associated with smoking. FDA apparently recognizes that underage smokers do not suffer from any lack of awareness of the risks of smoking. The Agency seeks to design new warnings that will 4 See page 70 of W. Kip Viscusi, Smoking: Making the Risky Decision New York: Oxford University Press, 1992. 5 See id. at 64. Alternative question wordings yielded stark results. In particular, results pertaining to assessments of the lung cancer mortality risk are similar to the assessed risk of lung cancer. More importantly, assessments of the overall smoking mortality risk are considerably greater than the assessed lung cancer risk. Whereas the assessed lung cancer risk was .43, the assessed total smoking mortality risk is .54. See id. at 77. Respondents view death from smoking as better than a 50/50 proposition. 6 See id. at 123. 7

! 0 • • "shout" at children, so to speak, to find ways not of imparting new information but of turning up the "volume" so that even the deafest youths will hear -- and act on -- the message. FDA's approach is reflected in the avowed intent of the focus group studies, which was to identify the message that the participants would find to be "most compelling. " 60 Fed. Reg. 61,670. The proposal by the youths who participated in the study that warnings might be given in bold letters, with neon print, and with jagged borders surrounding the warning, see 60 Fed. Reg. 61,674-75, is a reflection of the misguided intent of the entire exercise. The fact that the youths in the focus groups favor these design elements does not mean that they will be successful in changing behavior. Rather, this is simply a design that the focus group participants would like to look at!7 Aesthetically pleasing designs do not make warnings effective at changing behavior where there is no information deficit or misperception to correct.8 Counterproductive Effects Not only would additional warnings fail to reduce underage smoking, they might also have the unintended consequence of increasing smoking rates or promoting other 7 Several focus group participants clearly were more concerned with the aesthetics of FDA's warning designs than their efficacy. See, e. g. , Transcript of Focus Group Studies, November 9, 1995, 6:00 p.m., San Francisco, CA, at 66 ("MODERATOR DAVIS: But how about something like, you know, a different shape? Like this? PARTICIPANT: Oh, my God. PARTICIPANT: You need arrows. PARTICIPANT: That's really cool. That is super cool. MODERATOR DAVIS: Is that -- PARTICIPANT: The box and every -- PARTICIPANT: Flash neon pink, too. "); Transcript of Focus Group Studies, October 31, 1995, 6:30 p.m., Houston, TX, at 81 ("PARTICIPANT: Make it pretty or something. "); Transcript of Focus Group Studies, October 31, 1995, 4:30 p.m., Houston, TX, at 82 ("PARTICIPANT: (Inaudible.) Looks boring, but colors make it more exciting. "). 8 Indeed, aesthetics may work against the intended effects of a warning. Young children, for example, often find the skull and crossbones poison symbol to be an ZZd attraction rather than a deterrent. See Schiender K. "Prevention of Accidental Poisoning O through Package and Label Design," Journal of Consumer Research Vol. 4, September Cse 1977, pp. 66-74. © ~ W -8- ~ h~

• • • • • forms of risk-taking. FDA apparently has not considered all of the implications of the information that its proposed warnings would communicate. As the remarks of some of the focus group participants suggest, teenagers may interpret these warnings in ways that FDA did not anticipate. In view of teenagers' overassessment of smoking mortality risks, one wonders what will be the effect of the proposed FDA warning: "About 1 out of 3 kids who become smokers will die from their smoking. " This estimate is in line with my own published assessments of the total smoking mortality risk to smokers of .18-.36, though it is at the high end of my range.9 However, what is at issue here is not the accuracy of the statement but rather its effect on smoking risk perceptions. Youths and adults currently believe that the risk from smoking is considerably rg_eater than a probability of 1 out of 3. Thus, providing the information that 1 out of 3 kids will die from smoking may in fact reduce risk perceptions (and possibly increase smoking rates).10 A possible counterargument may be that at least this warning makes the information more precise and consequently more credible than would a possibly ambiguous subjective risk judgment. However, the role of ambiguity is a topic for which I have undertaken considerable research for EPA, particularly with respect to risks of cancer and other ambiguous risks regulated by that agency.11 In situations in which the 9 See page 70 of W. Kip Viscusi (1992). 10 Indeed, some focus group participants believed that some youths would interpret FDA's "About 1 out of 3 kids ..." warning to mean that two out of three kids could smoke without suffering increased mortality. See, e.Q., Transcript of Focus Group Studies, October 25, 1995, 6:30 p.m., Charlotte, NC, at 37 ("PARTICIPANT: Well, they' ll think, it won't happen to me, I' ll go ahead and do it anyway. They' ll think they' 1l be one of the two. PARTICIPANT: Yeah. "). 11 See, among my studies for EPA, W. Kip Viscusi, Wesley A. Magat, and Joel Huber, Communication of Ambiguous Risk Information," Theorv and Decision, Vol. 31, Nos. 2/3 (Sept./Nov. 1991) pp. 159-173 and also W. Kip Viscusi and Wesley A. Magat, "Bayesian Decisions with Ambiguous Belief Aversion, " Journal of Risk and Uncertainty, Vol. 5, No. 4 (Oct. 1992) pp. 371-387. 9

a • • • • • • • risk is ambiguous, the evidence shows that individuals are actually more reluctant to engage in the risky behavior than in situations in which the risk is precisely known. Thus, if the warning sharpens risk perceptions it will also encourage smoking rather than deter it. In each case the consequences are opposite those that FDA intends. If the warning indicates that one-third of all kids will die from smoking, how will adults assess the risk given this information? Since their period of exposure to the risk from cigarettes will be less than that for youths, will they assess their risk as being less than one-third? Because the adult population cannot be excluded from receiving a warning message that is included in cigarette advertising, one needs to be concerned not only with how the information will be processed by children but how it will be processed by adults as well. In the case of adults, the potential for erroneous interpretations with respect to the risk level for adults is particularly great. FDA must consider the effect of the warning on every potential smoking group that will be exposed to the warning message through advertising. Diminishing risk perceptions is not the only potentially counterproductive effect of FDA's contemplated actions. The proposed warning in which the risks of smoking are compared to a variety of other societal risks, such as those posed by AIDS and drunk driving, will tend to trivialize these risks in the view of the recipient of the warning. Providing comparative risk information always poses inherent difficulties, in part because it is predicated on the assumption that the reference point risk comparisons are well understood. If the respondent believes that the risks from smoking are not that great, will this then lead the recipient of the warning to dismiss other less well understood risks, such ` as -that of AIDS? Even if the respondent believes that the risks from smoking are great, the message that those risks are great compared to the risks of AIDS, etc., necessarily ~ 0 c.rc ~ ~ -10- ~ ci,D W