Tobacco Documents Online

726 STATEMENT OF PROFESSOR YO RAM J. WIND FOR SU$MISSON TO THE S'J3COiC4iTT_T? ON :Ei+LT:: AND TI•iF. S*'VIRODIiVF.`IT I am Yoram (Jerry) Wind. Since 1973 I have served as Professor of Marketing at the Wharton School of the Uni•rersit~ of Pennsylvania, and am the past editor of the Journal of Market- inc. I have been on the faculty at Wharton since receiving my ~ doctorate degree from Stanford University in 1a _67.' My spec<_al_,.y•i is marketing research, with particular emphasis on the ar.aiysi.s : and measurement of consumer behavior. During the past fifteen :. years I have served as a research consultant for various government agencies and about 100 companies and have published ~ extensively '-n many areas of marketing. A resume of my educa- r~ tional background and professional activities, and a bibliograp:i. of my publications, are attached to this statement. I have been asked by The Tobacco Institute to presentl my views, as an expert in marketing and consumer behavior, on y., ~ the theoretical and factual support for certain of the provisions of H.R. 5653. My testimony will deal with the proposed fir_diags1 that existing government and private programs, including the Surgeon General's warning statement, have not adecuately i._ formed the public about smoking and health issues, and the proposal to replace the current warning statement wi th a rotational system of seven different warning statements. My comments are based on an evaluation of a doc•.:ment issued in May 1981 by the Staff of the Federal Trade Commission e.*.t'_tled Report on the Cigarette advert' si ng Invest' gar.ion, •.rhic: I understand was submitted rationale that appears tc :abelinq provisions of tf the consumer studies and .:Pon in that Report. My conclusion c _x_ent that the labeling _.. :. R. 5653 are based or set forth in the FTC Sta; ,actual or theoretical st _:11's objective. I base that cor First, the leve sr:oking and health issue: a'.:rveys cited in the FTC b=S ~ -- much higher than 'crnal advertising and ma =:C Staff to minimize the :ssues are based on misir --ted in the Report and e ress and belief. The cor e t-.1 that "present Federe •`•a7e been insufficient ir A-erican Public" (Sec. 2( ,,,•- ~- • ngs of the studies : Second, there : support for She proposit: G7CO., Q: C ~ N ~ N]

727 A*d J. -WIND FOR 2 1 "_ . ON :~'AL:: ;]5:IT ince 1973 I have served -ton School of the University :or of the Journal of Ma:ket- narton since receiving my -sity in 1967.' My specialty r emphasis on the ar.alys9.s During the past fifteen :sultant for various :oanies and have published ig. A resume of my educa- :tivities, and a bibliography ::zis statement. 3bacco Institute to present and consumer behavior, on °or certain of the provisions 1 with the proposed findings programs, including the have not adequately i._ iealth issues, and the ing statement with a warning statements. My of a document :ssued in Trade Commission entitled Investigation, which I t understand was submitted to Congress and was based on a rationale that appears to be similar to that underlying the labeling provisions of the present bill. I also have reviewed the consumer studies and sur~zeys that are principally relied upon in that Report. . My conclusion can be stated in one sentence: to the extent that the labeling and advertising proposals contained in H.R. 5653 are based on the recommendations and cor.clusicns set forth in the FTC Staff Report, those proposals are without factual or theoretical support and are unlikely to achieve the bill's objective. I base that conclusion on two key points: First, the level of public awareness about various smoking and health issues, as demonstrated by the studies and surveys cited in the FTC Staff Report, is extraordinarily high -- much higher than one would expect to result from normal advertising and marketing me thods. The efforts of the FTC Staff to minimize the extent of nublic awareness or. these issues are based on misinterpretation and misuse of the studies cited in the Report and a fundamental confusion between aware- ness and belief. The conclusion presented in the proposed bill that "present Federal, State, and private initiatives have been insufficient in conveying the health messages to lc~':e American ?ublic" (Sec. 2(7)), is not consistent with the findings of the studies relied upon by the FTC Staff Report. Second, there is neither theoretical nor empirical support for the proposition that the rotational warr.ing system 0 CJ ~ Go N 4 GD ®

728 - 3 - proposed in H.R. 5653 and recommended by the FTC Staff would have any positive impact on the level of public awareness about ,, smoking and health issues. Replacement of the current warning' statement with seven different rotational warnings thus would be totally arbitrary. ~ I would like to elaborate on each of these points. - r With respect to the existing level of public aware- ness, the FTC Report begins its analysis with the ad.:.ission that "most people are generally aware"eof the claims about smoking and health. The report cites a 1978 Gallup Opinion poll, which indicates that more than 90 percent of the public believes that smoking is hazardous to health. Similar high percentages respond affirmatively to more specific issues: over 90 percent of the public believes that heart disease has been found to be associated with smoking; almost 90 percent believes that smoking during pregnancy can affect the smoker' baby; 87 percent of adults are of the view that smoking has been found to be associated with cancer of the mouth and with chronic bronchitis; well over 80 percent either "think" or . "know" that smokers are many more times as likely to develop lung cancer as nonsmokers. These responses are remarkable. National su rveys and polls consistently identify substantial segments of c:.e American public who are unaware of major public issues and z :icts -- the ener c t-her public lead A measu s:dered "deficien re:fect awareness ~ard is both theo =_:%.ts of human c -,eczanisms insure _e aware of or in :s why there is a 7a:ticularly a te. $-cz as the studiE An equa_ ='-::.camental misin' s=:=dies on which '_ =e _dentified. First, t ~~ec:fic question_ '=a== assumed thrc ~=a: if a number c '=out the smoking ==3t smoking durir. --= = and miscarri '''nera1 propositic =•e assertion that "s{s of adverse e

y the FTC Staff would ` f public awareness about' of the current uarr.ing al warnings thus would sach of these points. :evel of public aware- s with the admission )f the claims about 1978 Gallup Opinion percent of the public :a1th. Similar high -e specific issues: .hat heart disease has ;; almost 90 percent :an affect the smoker's .ew that smokir.g has of the mouth and with : either "think" or as likely to develop National surveys ial seyzments of the public issues and fl 729 - 4 - facts -- the energy crisis, the identity of the President and other public leaders -- the examples are numerous. A measured level of 90 percent awareness can be con- sidered "deficient" only if it is compared to a standard of perfect awareness. But it should be obvious that such a stan- dard is both theoretically and practically impossible. The limits of human cognitive abilities and selective perception mechanisms insure that 100 percent of any group wi11 never be aware of or in agreement about any fact or issue. TY:at is why there is a distribution of responses in any test, ~ particularly a test involving multipLe choice questions such as the studies cited in the FTC Staff ReDort. An equally important defect is the FTC Staf='s fundamental misinterpretation of the results of the consumer studies on which it relies. Six major misir.terpretations can be identified. First, the Staff improperly focussed on responses to specific questions, rather than on aatterns of responses. The Staff assumed throughout its discussion of public awareness ,hat if a number of people are not aware of a specific detail about the smoking and health issue -- for example, the claim that smoking during pregnancy increases the risk of still birth and miscarriage -- those people are not aware of the general proposition that encompasses that detail -- that is, the assertion that smoking during pregnancy increases the risks of adverse effects on the baby. 8 I 0 a

730 - 5 - This assumption is contrary to the actual results of the studies cited in the report, which show that most people are aware of all of the significant claims about smoking and health. It also violates fundamental principles about measure- ment of knowledge or awareness, which call for the development of an overall knowledge score or scores based or. response to multiple items. Can your knowledge of a subject, let's say economics, politics or health, be assessed accurately by your response to a single question on each topic? Yet this is analogous to the FTC Staff's reliance on the response to a single question on the health effects of sr4okir.g. The second area of misinterpretation is the Staff's assumption that anyone who responds incorrectly to multiple choice questions involving detailed statistics or medical knowledge is insufficiently aware of the fact or issue in- volved in the question. Respondents were asked numerous - questions involving precise details, such as "Out of every hundred people who get lung cancer, how many die from it," and were provided with six alternative answers. According to the Staff, anyone who picked other than the answer that the Staff called correct -- 95 -- "did not appreciate the severity of lung cancer." In fact, however, the vast majority of respondents chose either 45, 75, or 95, indicating that they believe that lung cancer has a high mortality rate of at least 1 of every 2 of those suffering from the disease. 2'hat belief hardly supports the Staff's conclusion. Similarly answers to such que ~:f the respondent. _o the interpretati, s_atement presented :avolved. In my prr, -ost people with luz ,;as not sure whethe: _7 out of 100, migh' erroneously have inc :o "do not appreciz Still anot answer is the one of ""•:ssissipDi: "I waE and I did. I said I .nterviewer's Manual -niversity of Michic_ no opinion on that" t.`a naing" and advise "don't know" respons studies relied upon Third, man to the studies could standing of the ques -•^n the Report on the 4 ;-077 0-82--47

to the actual results of : show that most people :aims about smoking and principles about measure- call-for the development :s based oa response to : a subject, let's say :ssed accurately by your topic? Yet this is on the respor.se to a of smoking. ,retation is the Staff's .correctly to multiple atistics or medical he fact or issue in- ere asked numerous uch as "Out of every w many die from it," answers. According than the answer that the appreciate the severity a vast majority of , indicating that they rtality rate of at least ie disease. 2^_iat belief - 6 - Similarly, the Staff interpreted "don't know" answers to such questions as a lack of awareness on the part of the respondent. But such an answer is equally susceptible to the interpretation that the respondent is aware of the statement presented but is unsure of the precise statistics involved. In my previous example, a person who believed that most people with lung cancer die from that disease, but who was not sure whether the correct proportion is 85, 90, 95, or 97 out of 100, might answer "don't know." The Report would erroneously have included that Aerson in the category of those who "do not appreciate the severity of lung cancer." Still another interpretation of the "don't know" answer is the one offered by Mark Twain in Life on the Mississippi: "I was gratified to be able to answer promptly, and I did. I said I didn't know." Recognizing this, the Interviewer's Manual of the Survey Research Center of the University of Michigan points out that the answer "I have no opinion on that" can mean merely "wait a minute, I am thinking" and advises that it is a good idea to probe all "don't know" responses. No probing was conducted in the studies relied upon by the FTC Staff. Third, many of the so-called "incorrect" responses to the studies could have resulted from simple lack of under- standing of the questions. The studies primarily relied upon in the Report on the issue of public awareness were based on 95-077 0-82--47 C W ~ C W N N N

732 M 0 - 7 - telephone surveys. In s•ich interviews, it is not reasonable to expect a high percentage of correct answers to complex questions such as those that were asked in the studies. Let me give you an example of one such question. Imagine that I have called you out of the blue, explained who I am, and ask you a long series of questions. Even if you have remained intere_ted and alert throughout the interview, you must respond to such questions ag this one: "How many Americans living today will eventually die from diseases related to smoking cigarettes? None, one out of two, one out of six, one out of ten, or one out of a hundred?" N The ambiguity of the phrasing -- does the word "Americans" r>fer to all Americans or only those who smoke? -- as well as the precise statistical answers presented turn such surveys into a guessing game rather than a test of knowledge or awareness. Fourth, none of the studies cited in the Report in- cluded supporting data as to the reliability and validity of their findings. That is, no evidence was offered to show that the series of questions asked were a reliable measure of public awareness or knowledge about any particular issue. For example, how many respondents would provide the same answers if they were reinterviewed a few weeks later? Furthermore, no validation of the results was provided. indeed, given the focus of the FTC Staff Report on substantive conclusions, i0 it were submitted to a professional publication such as the J~urnal of Marketing today, it would be re procedures for the s•. Fifth, neit presented any norm ac .a the classic text c w'_11iam Angoff of thc "By now it raw scores they are ac data that , interpreti~ In other wc that the answers to z `:igh, 1ow or medium : standard that would : Sixth, and sions, and to a larg, clusions are based, : distinct concepts of : e Staff Report def. ". . . tho statements estimate o: ansWer~doi Staff Repo- '-^-:ere is no eonceptuu diverse responses. -ooki:g for exact re

~iews, it is not reasonable :rect answers to comclex asked in the studies. 3le of one such question. = of the blue, explair.ed who ` questions. Even if you : throughout the interview, 3 ag this one: ing today will eventually die zigarettes? None, one out of :n, or one out of a hundred?" rasing -- does the word s or only those who smoke? -- 1 answers presented turn such er than a test of knowledge ar._. _.-. . ._.. .. .. _ , . dies cited in the Report in- reliability and validity of .ence was offered to show !d were a reliable measure of :ut any particular issue. For ad provide the same answers weeks later? Furthermore, provided. indeed, given the substantive conclusions, if :al publication such as the 733 -a- Journal of Marketing during my editorship, or Marketing Science today, it would be rejected due to the lack of any validation procedures for the studies upon which its conclusions are based. Fifth, neither the Staff Report nor the studies presented any norm against which to compare the survey results. In the classic text on testing, Educational Measurement, William Angoff of the Educational Testir.g Service states: "By now it has become almost axiomatic that raw scores on a test yield no meaning unless they are accompanied by relevant supplementary data that will place the score in an appropriate interpretive context." In other words, it is impossible validly to conclude that the answers to a particular survey item represents a high, low or medium level of awareness in the absence of a standard that would show what answers are to be expecced. Sixth, and most fundamentally, the Report's conclu- sions, and to a large extent the studies upon which those con- clusions are based, reflect a hopeless confusion of the very distinct concepts of awareness versus knowledge and belief. The Staff Report defines "unaware" as: ". .. those who say they do not believe true statements ; believe false statements; under- estimate on a multiple choice question or answer ~don't know' or uncertain." (FTC Staff Report at p. 17 note b, emphasis added) There is no conceptual justification for combining these five diverse responses. Belief is not awareness or !ciowledge; looking for exact response an a multiple cho:ce question in M I

734 - 9 - search of the precise "correct" answer is not reasonable, "don't know" is both a function of the a_uestion wording and the item of concern; and uncertainty does not necessarily represent lack of awareness. One example of this faulty definition of "awareness" is the FTC Staff's misuse of the 19c0 Roper Study cited in the Report. That study asked people "how true you personally think" a particular statement is. The interviewer was in- structed to introduce those questions to the respcr_dents as follows: "Now I'm going to read you some statements about smoking and health, and for each one I'd like you toNtell me your beliefs about how true the statement is." Respcndents were permitted to answer only "know it's true," "think it's true," "don't know if it's true," "think it's not true," or "know it's not true." The Staff concluded that those respcr.d- ing in the latter three categories are "unaware" of the ir.for:aa=_ tion conveyed by the specific statements made. It is obvious, however, that a respondent could be aware of a claim (for example, that smoking increases the risk of heart attack) yet disagree with it. The FTC Staff consistently misused the studies in this respect to support its erroneous conclusions about awareness. Let me read another example from page 3-i9 of the Report: "According to the Gallup Opinion Index, June, 1978, 19% of the population do not believe that smoking causes lung cancer . . . . Among all smokers, 28% did not believe smoking caused Lung cancer while amcng heavi er smokers, nearly one-third -- 31% -- did not believe or 7 - 10 know about the link. wide, these data sugg of Americans, both sm do not know that ciga cancer.°- =:e Staff's conclusion simply d ._s fundamental type of error consumer belief -- undermi r.:ght be based on a premise tha _-:formed about smoking and heal The FTC Staff Report .,:'_1 -- proceed from the premis :nformed about smoking and heal t:at a new system of warning st =--tify this inadequacy. As I ._e available data do not supnc e.c-,saLLy disturbing that the neG ~ - recommended by the Report and t achieve any positive impact on s-oking and health issues. There is no evidence those proposed in H.R. 5653 wot =%~hiic awareness. The only stt relates to this issue is a stuc =y 3urke Marketing Research to ty?es of proposed warning stat,

1er is not reasonable, :he a_uestion wording and ! does not necessarily ! definition of "awareness" i0 Roper Study cited in : "how true you personally :he interviewer was in- :s to the res-zcr.denzs as : some statements about ! I'd like you to tell statement is." Respondents it's true," "think it's :hink it's nat true," or :cluded that those respcr_d- .re "unaware" of the ir.forma- :ents made.`•It is obvious, aware of a claim (for risk of heart attack) misused the studies i.^n this .clusions about awareness. -e 3-19 of the Report: pi nion Index, June, n do not bel=eve ancer . . ?.mor_g elieve smoking mcnq heavier smokers, did not believe or 735 know about the link. . . Projected r.ation- wide, these data suggest that tens of millions of Americans, both smokers and r.on-s-:okers, do not know that cigarette smcking causes lung cancer. " The Staff's conclusion simply does not follow from the data. This fundamental type of error -- confusing consumer awareness with consumer belief -- undermines any recor,urnendations that might be based on a premise that the public is not adequately informed about smoking and health. The FTC Staff Report -- and presumably the present bill -- proceed from the premise that the public is inadequately informed about smoking and health issues to the cor.clusion that a new system of warning statements is necessary to rectify this inadequacy. As I have attempted to demenstrate, the available data do not support the premise. 3ut it is equally disturbing that the new system of warning statements ~ recommended by the Report and by E.R. 5653 are not likely to achieve any positive impact on consumer awareness about smoking and health issues. There is no evidence that specif•:c warnings such as those proposed in H.R. 5653 would result in any increase in public awareness. The only study of which I am aware that relates to this issue is a study conducted for the :TC Staff by Burke Marketing Research to test the recall of different types of proposed warning statements and :or*sats. This study