- Email: [email protected]
Effects of Sales Promotion on Smoking among U.S. Ninth Graders
Preventive Medicine 28, 243–250 (1999) Article ID pmed.1998.0410, available online at http://www.idealibrary.com on
Effects of Sales Promotion on Smoking among U.S. Ninth Graders William H. Redmond, Ph.D.1 College of Business Administration, Bowling Green State University, Bowling Green, Ohio 43403
INTRODUCTION
Objective. The purpose of this study was to examine the association between tobacco marketing efforts and daily cigarette smoking by adolescents. Design. This was a longitudinal study of uptake of smoking on a daily basis with smoking data from the Monitoring the Future project. Diffusion modeling was used to generate expected rates of daily smoking initiation, which were compared with actual rates. Study data were from a national survey, administered annually from 1978 through 1995. Between 4,416 and 6,099 high school seniors participated per year, for a total of 94,652. The main outcome measure was a deviation score based on expected rates from diffusion modeling vs actual rates of initiation of daily use of cigarettes by ninth graders. Annual data on cigarette marketing expenditures were reported by the Federal Trade Commission. Results. The deviation scores of expected vs actual rates of smoking initiation for ninth graders were correlated with annual changes in marketing expenditures. The correlation between sales promotion expenditures and the deviation score in daily smoking initiation was large (r 5 0.769) and statistically significant (P 5 0.009) in the 1983–1992 period. Correlations between sales promotion and smoking initiation were not statistically significant in 1978–1982. Correlations between advertising expenditures and smoking initiation were not significant in either period. Conclusions. In years of high promotional expenditures, the rate of daily smoking initiation among ninth graders was higher than expected from diffusion model predictions. Large promotional pushes by cigarette marketers in the 1980s and 1990s appear to be linked with increased levels of daily smoking initiation among ninth graders. q1999 American Health Foundation and Academic Press Key Words: cigarette smoking; adolescent smoking; tobacco marketing; sales promotion; diffusion.
1
Fax: (419) 372-8062. E-mail:[email protected].
The uptake of cigarette smoking among adolescents follows a regular pattern that is well known to health researchers. In particular, first trial of cigarettes rises throughout elementary school, peaks around the 6th grade, and declines annually thereafter [1,2]. The incidence of smoking on a daily basis lags first trial, usually peaking in the 9th grade. This basic pattern has been observed for successive cohorts for many years; that is, it represents a highly stable phenomenon of the teenage years that recurs for each class. Data on grade in which initiation of daily smoking occurs show a marked peak in 9th grade. Ninth grade rates of daily smoking initiation are, on average, 40% higher than 8th-grade and 20% higher than 10th-grade levels, which are in turn higher than 7th- or 11th-grade levels. In the data examined below, 9th grade was the peak year for 15 of the 16 cohorts examined, while the 9th-grade rate in the remaining cohort was near to the peak figure (6.4 vs 6.9% of the class). This pattern reflects a growing propensity to regularize smoking as the cohort progresses through the 9th grade, followed by reduced propensity after the ninth grade. After 9th grade the attractiveness of smoking appears to progressively decline, as the resolve of nonsmokers to remain nonsmokers strengthens. In other words, 9th-grade represents the point of maximum hazard for adolescents becoming daily smokers. A nonsmoker in 9th grade who reaches the 10th grade in that condition is at decreasing risk of becoming a smoker thereafter. Because of this pattern, the 9th-grade population contains a larger proportion than other grades of individuals who are in an intermediate state of inclination with respect to smoking. That is, the 9th grade has a high proportion of individuals lacking either a strong predisposition to start smoking or a strong commitment to avoid smoking. Compared with others having stronger inclinations, these students are expected to be more sensitive to outside influence in their choice of smoking or nonsmoking. The present research examines 9th
243
0091-7435/99 $30.00 Copyright q 1999 by American Health Foundation and Academic Press All rights of reproduction in any form reserved.
244
WILLIAM H. REDMOND
graders precisely because influences on smoking behavior are likely to be more readily detectable. Influences impinging on these “fence sitters” may come from a variety of sources, including friends, entertainers, parents, teachers, or marketers. The present research focuses on marketing influences, with particular attention to sales promotion expenditures. SALES PROMOTION
The influence of tobacco-marketing activities on adolescents has been the subject of discussion and debate for a considerable time. In both public policy forums and the tobacco control literature, the focus has been mainly on the role of advertising. However, a growing body of research shows that a substantial number of adolescents are affected by tobacco promotions [3–5]. Concerns about the effects of promotions on youths are sufficiently strong that the Food and Drug Administration’s proposed rules regulating tobacco marketing have the effect of prohibiting or restricting many of the promotional tactics in current use by tobacco companies, in addition to advertising restrictions [6]. As is the case with many consumer products, advertising and promotion are used in tandem in cigarette marketing [1,3]. Although advertising and promotion have the same ultimate goal of increasing sales, their roles are distinct [7,8]. Advertising is meant to influence perceptual factors such as brand awareness, image, and positioning. That is, the purpose of advertising is to condition the consumer’s perceptual structure in a way that is both positive and enduring. Sales promotion, on the other hand, is more closely linked to actions than perceptions. Specifically it is meant to generate trial of the product or influence selection of the brand over other brands. Sales promotion is aimed at stimulating behaviors rather than conditioning perceptions and is expected to have more of a near-term impact than a long-run effect. Sales promotion tactics include coupons, free samples, premiums, in-store displays, and a variety of other methods and are notable for visibility and for presence at the point of purchase. Unlike the marketers of most other consumer products, however, tobacco companies do not advertise on radio and television, so that their reliance on promotional activities is comparatively greater. In fact, expenditures for sales promotion of cigarettes have grown to be substantially higher than advertising expenditures in the United States. Figure 1 illustrates trends in advertising and sales promotion after 1970, which was the last year of radio and television advertising for cigarettes (data are as reported to the Federal Trade Commission, unadjusted for inflation). Because of their heavy reliance on sales promotion,
cigarette marketers employ a very wide range of promotional tactics [2]. These include, in descending order of expenditures, coupons and special offers, promotional allowances to distributors, speciality items (e.g., logo merchandise), point-of-purchase displays, public entertainment (e.g., sports event sponsorship), samples, direct mailing, and audiovisuals. In terms of consumer reaction, the categories of coupons, special offers, and samples have the effect of lowering the effective price of smoking, while promotional allowances and point-ofpurchase displays have the effects of wider availability, favorable shelf location, and enhanced in-store visibility. Sponsorships serve to keep the brand name visible and linked to desirable activities; specialty items and direct mail also keep the brand visible and frequently link the product to peers and reference groups. Cigarette marketing campaigns may or may not be intentionally aimed at adolescents, but it is clear that such efforts are seen by youths and are attractive to them: survey evidence shows high rates of participation in tobacco promotions by adolescents. Over one-third of minors reported some level of participation, with about 1 in 10 reporting possession of one or more promotional items [4]. Willingness to use a promotion item exceeds 25% among junior high and high school-aged teenagers [3]. Awareness of and participation in tobacco promotional campaigns has been found to be significantly associated with susceptibility to tobacco use, whether examined alone or in conjunction with advertising effects [9,10]. In sum, the evidence shows that youths are attracted by tobacco promotions and youths with a higher degree of involvement are more susceptible to smoking. UPTAKE OF SMOKING
The process of smoking initiation among adolescents has been adequately examined elsewhere and is discussed here only in a limited sense, specifically to establish a methodological linkage to diffusion. A number of researchers have observed that the spread of smoking among a cohort of adolescents bears similarities to the spreading acceptance of a new product in a market. That is, the spread of smoking is a diffusion process [11,12]. While cigarettes have been on the market for a considerable time, they are a new experience for each new group of adolescents, and the adolescents react to these as such. Similarities between the uptake of smoking and the spreading acceptance of new products are multiple. Prospective smokers recognize both advantages and drawbacks of smoking and a portion choose to adopt because of the perceived advantages while others are deterred by a focus on the drawbacks. Prospective smokers are influenced by multiple sources, including the actions and opinions of peers, role models, and marketers [13].
EFFECTS OF TOBACCO PROMOTION ON NINTH GRADERS
245
FIG. 1. Advertising and sales promotion expenditures: 1971–1994. Source: Federal Trade Commission.
Also, the spread of smoking among a cohort of adolescents has contagion-like properties, a hallmark of the diffusion process [14]. A small proportion of students begin smoking on a regular basis in elementary school; more subsequently follow the example of these early smokers and, in turn, influence the uptake of smoking by still other students. As noted above, the diffusion of smoking among adolescents has been found to be a highly predictable phenomenon. Experimenting with cigarettes starts at an earlier age than frequent smoking, with the peak time of first trial at sixth or seventh grade [15]. Recently, about 2 or 3% of the class have become daily smokers by the sixth grade, with an increasing rate through the ninth grade, then a decreasing rate through high school and beyond. The main difference for cohorts prior to the mid-1980s is that the sixth-grade rate was closer to 5% and a higher overall percentage became smokers. However, the basic pattern of diffusion was the same, with peak rates of initiation of daily smoking in the ninth grade. The methodological link consists in the use of diffusion modeling techniques to quantitatively characterize the spread of smoking in each of several cohorts of teenagers. Diffusion models of the kind used here have become standard techniques for analyzing and forecasting the diffusion of new products [16]. The technique has also been applied to a number of nonproduct situations, including the spread of smoking cessation among adults [17]. For the present research, diffusion parameter estimates are used to establish a baseline against which actual smoking initiation is compared. In other words, diffusion modeling furnishes a set of expected values for each cohort with which to compare actual
smoking data. Because each cohort can and does have slightly different rates of diffusion, the modeling approach employed here has the advantage of being a more sensitive indicator than alternative methods such as cross-cohort averages or extrapolations. DESIGN, DATA, AND METHODS
Design The study is designed to assess the impact of marketing activities on daily smoking initiation of ninth graders by making two types of comparisons. First is the influence of advertising compared with sales promotion. Second is a comparison of two time periods in which the basic characteristics of marketing strategy changed. The design is responsive to calls for research on smoking that utilizes large samples of adolescent smokers and nonsmokers, that analyzes longitudinal data, and that employs advanced statistical models [1,18]. Eighteen cohorts were separated into two groups based on their exposure to different regimes of marketing strategy. Eight of these are from the period in which sales promotion was less than advertising; these were ninth graders in 1978–1982. Ten cohorts are from the period in which sales promotion exceeded advertising; these were ninth graders in 1983–1992. For each of the 18 cohorts, a separate diffusion model was estimated from survey returns of daily smoking initiation in grades 6 through 12. The diffusion model established a “normal” or expected rate of smoking initiation for the ninth grade. The quantity of interest is the difference between the expected rate and the actual rate of initiation in the ninth grade. Thus the main criterion variable
246
WILLIAM H. REDMOND
is a deviation score representing the extent to which daily smoking initiation was higher or lower than the level that would be expected if the diffusion process was operating normally. Figure 2 shows actual and expected values for one of the cohorts. For the cohorts in each group, the degree of association between advertising changes, promotional changes, and deviation score was assessed by correlation. Data Data for daily smoking initiation are from the University of Michigan’s Monitoring the Future project, which examines a wide variety of substance use behaviors and related attitudes among high school seniors [e.g., 19]. The question used in the present research asks respondents to report the grade in which they first began to smoke cigarettes on a daily basis. This question is item 5570 and appears on two of the five survey forms (in the 1992 returns, these are questions B100A of Form 1 and E10A of Form 3). Sample sizes responding to this question varied by year, from 4,416 to 6,099, for a total of 94,652 in the 18 cohorts. Daily use of cigarettes is associated with the daily nicotine addiction cycle [2]. That is, this measure differs from a first-use type of measure in that it reflects regular use and dependence rather than occasional use or experimenting with cigarettes. Data on cigarette marketing expenditures are reported by the Federal Trade Commission [1,20]. Advertising expenditures include expenditures for newspaper, magazine, billboard, and public transit placements.
Promotional expenses include the categories described above. The data used for this research were year-toyear changes in expenditures, expressed as a percentage. Data reported for 1991 and 1992 were adjusted to correct for the effects of a change in the pricing strategy of cigarette companies [21]. This situation involved the reaction of “name brand” cigarette marketers to the market share erosion caused by generics and private label cigarettes [22]. The price strategy of name brands was to increase their coupon activity to the degree that it represented a broad, but temporary, price reduction. Finding this less than fully satisfactory, the name brand marketers switched to a permanent price cut in 1993, at which time overall industry coupon expenditures fell back to their previous levels [23,24]. The data adjustment was to fix coupon expenditures for 1991 and 1992 at levels that prevailed in 1990 and 1994; other categories were not affected. That is, the purpose was to remove a transient pricing strategy from the promotional classification. Model The diffusion framework used to evaluate the data is the Bass [25] diffusion model, which is the standard methodology for assessing diffusion in marketing [14,16]. This specific model is well known in the literature on new product diffusion; however, the basic model form is closely related to contagion modeling in epidemiology. In this model, the number of new adopters in period t, (St), is a function of the eventual number of adopters in the cohort (m), the number of individuals
FIG. 2. Actual vs expected daily smoking initiation for the high school class of 1988.
EFFECTS OF TOBACCO PROMOTION ON NINTH GRADERS
who have already adopted (Yt21), and the probability that a nonadopter adopts in period t (Pt),
247
prevalent in marketing campaigns, the results of the present study may understate the effects of advertising and/or promotion on teenage smoking.
St 5 (m 2 Yt21)Pt , RESULTS
where Pt 5 p 1 q [Yt21 /m]. Here, p is the innovation parameter and q is the imitation parameter. The innovation parameter represents the earliest starters, while q represents the basic diffusion rate, that is, the rate of spread among the main group of potential adopters. Parameters of the model were estimated using the maximum likelihood procedure outlined by Takada and Jain [26] on SAS NLIN software. Limitations Interpretation of the results is made within the following limitations. Self-reported measures of smoking have been found to be accurate when checked against chemical evidence [27]. However, the data on smoking initiation were obtained in a recall format and may therefore be subject to inaccurate recollection as to the grade of initiation [15]. Also, some respondents may have started daily smoking in the interim between one grade and the next, and made an arbitrary assignment as to grade. For these kinds of errors, the magnitude is uncertain but is thought not to be large, and there is no indication of a systematic or directional bias. A second limitation is that the smoking initiation data have a directional bias; however, the magnitude of this error is well understood. This is a result of surveying 12th graders thereby excluding high school dropouts, who are known to exhibit relatively higher smoking rates and earlier initiation times than high school graduates. Had dropouts been included in the survey, initiation rates for all grades except the 12th would be higher, with the 9th-grade rate 4.0–4.5% higher [15]. Such an inclusion would result in modest increases in model parameters p and m. Absent evidence that dropouts react in a systematically different way to marketing appeals, there is no basis upon which to anticipate substantively altered correlation results. Last, because the advertising and sales promotion data are based on expenditures, these measures reflect the likelihood of exposure. In other words, the measures do not account for qualitative aspects of exposure and therefore are not sensitive to the presence of marketing campaigns that are either especially appealing to youths or off-putting to youths. Previous studies indicate that certain types of content are associated with higher brand awareness and market share among young smokers [28]. To the extent that such content is
Diffusion Modeling Model diagnostics indicate a good fit of the smoking data with the basic diffusion process of increasing rates of initiation followed by decreasing rates. Goodness of fit, as measured by R 2, ranged between 0.92 and 0.99 for the 18 models, which is slightly above average values for diffusion models based on new consumer products [29]. Across time, parameter estimates reflected broad trends in teenage smoking, with an overall decline in smoking rates through the 1980s and the start of an increase in the early 1990s. Among the 18 cohort models, estimates for the innovation parameter ( p) ranged between 0.01 and 0.05; estimates for long-run daily smoking potential (m) ranged between 28.9 and 50.4. The estimates for longrun potential can be interpreted as percentages of the cohort: in the earliest cohort, slightly over 50% were estimated to become daily smokers at some point in their lives. Estimates for the imitation parameter (q) below are rates and are not interpretable as percentages of the cohort. Estimates for the imitation parameter (q) ranged between 0.612 and 0.771. The imitation parameter (q) is frequently a main focus of interest in diffusion studies because it reflects the speed with which a new product spreads within a group. For comparative purposes, the q parameter for new products in the United States averages about 0.30 [30]. In contrast, the q parameter for smoking cessation among adults averages about 0.19 [17]. Compared with the diffusion of new consumer products generally, the uptake of smoking among adolescents is a relatively rapid process, while quitting smoking is a relatively slow process. The criterion variable is a deviation score of expected vs actual smoking. This is computed by subtracting the model prediction of percentage initiating daily smoking in the ninth grade from the actual percentage initiating daily smoking in the ninth grade. This quantity was positive for all years, reflecting the sharp peak that characterizes uptake of daily smoking in the ninth grade and ranged between 10.25% and 11.48%. These data are shown in Fig. 3. The deviation scores are higher on average in the 1983–1992 period than in the 1975– 1982 period (10.66% vs 1 0.57%) and the standard deviation is also larger (0.31 vs 0.22); however, differences between the two periods are not statistically significant at P 5 0.10 or better.
248
WILLIAM H. REDMOND
FIG. 3. Ninth-grade smoking deviation scores by year.
Correlations Associations between marketing activities and daily smoking were estimated by standard Pearson productmoment correlation techniques. Results are shown in Table 1. For the 1975–1982 period, the correlation coefficient between advertising expenditures and the smoking deviation score is larger than the coefficient between sales promotion expenditures and the smoking deviation score, although neither is significant at P 5 0.10 or better. For the 1983–1992 period, the correlation coefficient between sales promotion expenditures and the smoking deviation score is larger than the coefficient between advertising and smoking and is statistically significant. This correlation coefficient is 0.769 and is significant at P 5 0.009. The correlation coefficient between smoking and advertising is not statistically significant in this time period. The relative strength of association between advertising and the smoking deviation score is reversed in the two time periods, which moved in the opposite direction of the association between promotion and the smoking deviation score. For the period in which sales promotion expenditures were lower than advertising (1975–1982), the association between promotional expenditures and the smoking deviation score is relatively weak r 5 0.287; P 5 0.49). TABLE 1 Correlation Coefficients Relating Deviation in Daily Smoking Initiation to Sales Promotion and to Advertising Expenses Correlation between smoking deviation score and
9th graders in 1975–1982
9th graders in 1983–1992
Sales promotion expenditures
0.287 (P 5 0.490) 0.534 (P 5 0.172)
0.769 (P 5 0.009) 0.246 (P 5 0.493)
Advertising expenditures
For the period in which sales promotion expenditures exceeded advertising (1983–1992), the association between sales promotion expenditures and the smoking deviation score is positive, substantial, and statistically significant. The magnitude of the correlation coefficient indicates that year-to-year changes in sales promotion expenditures account for about 54% of the variance in the smoking deviation score. CONCLUSION
Prior research has established that cigarette sales promotions are attractive to adolescents and that receptivity to promotional items by adolescents is associated with progression toward smoking behavior [3,5]. The present research establishes a linkage between cigarette promotions and escalated levels of smoking behavior among adolescents. Significantly greater numbers of ninth graders initiate daily smoking in years of high promotional expenditures than would otherwise be expected. The shift by tobacco marketers from an emphasis on advertising to an emphasis on sales promotion appears to have altered the nature of ninth graders’ response to marketing efforts. Since the early 1980s, year-toyear changes in sales promotional expenditures for cigarettes have been associated with deviations in the percentage of ninth graders who start smoking on a daily basis. Previously, such linkages were not statistically significant. The role of sales promotion as a behavioral trigger is manifest in the data only after the point in time at which promotional expenditures came to dominate the marketing strategy of tobacco companies. Whether or not aimed at adolescents, promotional activities are linked to elevated levels of daily smoking in that age group. The pronounced upward trend in sales promotion expenditures by cigarette companies is accompanied by several instances of large promotional “pushes” in which year-to-year increases of $400 million
EFFECTS OF TOBACCO PROMOTION ON NINTH GRADERS
or more are observed. The magnitude of effects is illustrated by figures from the year 1985. This year was the largest percentage increase of the 1980s: promotional expenditures in 1985 rose by $479 million over the 1984 level of $1.0 billion. In 1985 the ninth-grade class had an excess in daily smoking initiation of 0.86% after correcting for the ninth-grade peak factor (the average ninth-grade deviation is 10.62% while the deviation in 1985 is 11.48%). Assuming a cohort size of approximately 3.4 million students, this excess initiation rate translates into roughly 29,000 daily smokers above the level that would otherwise be expected. Further research will be required to ascertain whether sales promotions have a differential impact on racial, ethnic, or gender subgroups or whether the various forms of sales promotion have a differential effect on adolescent smoking. Design Considerations By design, this research is sensitive to deviations in rates of daily smoking initiation, rather than in base rates of daily smoking initiation. It is therefore possible that sales promotion also plays a more fundamental role in daily smoking uptake, but one which is undetectable in the present design. That is, it may be that promotional activities contribute to an overall predisposition to smoke and have, as a result, elevated the “expected” rate of daily smoking above rates that would be observed in the absence of large promotional campaigns. As noted above, the design precludes an evaluation of the qualitative aspects of promotional activities, as well as nonmarketing sources of influence on smoking behaviors. Lack of significance of the advertising variable does not necessarily imply an absence of causative links between advertising and daily smoking initiation. Advertising is expected mainly to have a long-term, cumulative effect in shaping attitudes and images. The present results indicate that year-to-year changes in advertising expenditure are not significantly associated with deviations in daily smoking initiation in that same year. The literature on adolescent smoking typically positions advertising as being among the determinants of predisposition to smoke, that is, one factor in a longterm accumulated inclination [10]. However, research on alcohol advertising indicates that geographic or temporal aggregation of data tend to obscure the linkages [31]. It is thus possible that a disaggregated approach may prove more insightful in cigarette advertising. ACKNOWLEDGMENT The author thanks Floris Wood and Darryl Stone for valued assistance in data retrieval.
249
REFERENCES 1. U.S. Department of Health and Human Services. Preventing tobacco use among young people: a report of the Surgeon General. Atlanta: U.S. Department of Health and Human Services, 1994. 2. Lynch B, Bonnie R, editors. Growing up tobacco free: preventing nicotine addiction in children and youths. Washington: National Academy Press, 1994. 3. Gilpin E, Pierce J, Rosbrook B. Are adolescents receptive to current sales promotion practices of the tobacco industry? Prev Med 1997;26:14–21. 4. Coeytaux R, Altman D, Slade J. Tobacco promotion in the hands of youth. Tobacco Control 1995;4:253–7. 5. Pierce J, Choi W, Gilpin EA, Farkas A, Berry L. Tobacco industry promotion of cigarettes and adolescent smoking. JAMA 1998;297:511–5. 6. Gostin L, Arno P, Brandt A. FDA regulations of tobacco advertising and youth smoking: historical, social and constitutional perspectives. JAMA 1997;277:410–8. 7. Kotler P. Marketing management: analysis, planning, implementation and control. 9th ed. Upper Saddle River (NJ): Prentice Hall, 1997. 8. Farris P, Quelch J. Advertising and promotion management: a manager’s guide to theory and practice. Radnor (PA): Chilton, 1983. 9. Altman D, Levine D, Coeytaux R, Slade J, Jaffe R. Tobacco promotion and susceptibility to tobacco use among adolescents aged 12 through 17 years in a nationally representative sample. Am J Public Health 1996;86:1590–3. 10. Evans N, Farkas A, Gilpin E, Berry C, Pierce J. Influence of tobacco marketing and exposure to smokers on adolescent susceptibility to smoking. J Nat Cancer Inst 1995;87:1538–45. 11. Pierce J. Progress and problems in international public health efforts to reduce tobacco usage. In: Omenn G, editor. Annual review of public health, Vol. 12. Palo Alto (CA): Annual Reviews, 1991:383–400. 12. Novak T, Bass F, Kamakira W, Dent C. Diffusion models for smoking onset and cessation: a segment-level historical analysis. In: McAlister L, Rothschild M, editors. Advances in consumer research, Vol. XX. Provo (UT): Association for Consumer Research, 1993. 13. Elders MJ, Perry C, Eriksen M, Giovino G. The report of the Surgeon General: preventing tobacco use among young people. Am J Public Health 1994;84:543–7. 14. Rogers E. Diffusion of innovations. 4th ed. New York: The Free Press, 1995. 15. Johnston L, O’Malley P, Bachman J. National survey results on drug use. Washington: U.S. Department of Health and Human Services, 1996. 16. Mahajan V, Mueller E, Bass F. New product diffusion models in marketing: a review and directions for research. J Market 1990;54:1–26. 17. Redmond W. Product disadoption: quitting smoking as a diffusion process. J Public Policy Market 1996;15:87–97. 18. Pollay R. Hacks, flacks, and counter-attacks: cigarette advertising, sponsored research, and controversies. J Social Issues 1997;53:53–74. 19. Bachman J, Johnston L, O’Malley P. Monitoring the future: questionnaire responses from the nation’s high school seniors. Ann Arbor (MI): Univ. of Michigan, 1993. 20. Federal Trade Commission. Report to Congress for 1990: pursuant to the Federal Cigarette Labeling and Advertising Act. Washington: Federal Trade Commission, 1992.
250
WILLIAM H. REDMOND
21. When smoke got in their eyes. Economist 1993;327:65–6. 22. Low-price smokes fire up promo flurry. Advertising Age 1990;61:33–4. 23. Phillip Morris making lower prices for Marlboro cigarettes permanent. New York Times 1993 Jul 21. 24. Add RJR to list of cig price cuts. Advertising Age 1993;64:3/46. 25. Bass F. A new product growth model for consumer durables. Manage Sci 1969;16:215–27. 26. Takada H, Jain D. Cross-national analysis of diffusion of consumer durable goods in Pacific rim countries. J Market 1991;55:48–54. 27. Patrick D, Choadle A, Thompson D, Diehr P, Koepsell T, Kinne
S. The validity of self-reported smoking: a review and metaanalysis. Am J Public Health 1994;84:1086–93. 28. Pollay R, Lavack A. The targeting of youths by cigarette marketers: archival evidence on trial. In: McAlister L, Rothschild M, editors. Advances in consumer research, Vol. XX. Provo (UT): Association for Consumer Research, 1993. 29. Wind Y, Mahajan V, Cardozo R, New product forecasting. Lexington (MA): Lexington, 1981. 30. Sultan F, Farley J, Lehmann D. A meta-analysis of applications of diffusion models. J Market Res 1990;27:70–7. 31. Saffer H. Alcohol advertising and motor vehicle fatalities. Rev Econ Stat 1997;79:431–42.
Effects of Sales Promotion on Smoking among U.S. Ninth Graders William H. Redmond, Ph.D.1 College of Business Administration, Bowling Green State University, Bowling Green, Ohio 43403
INTRODUCTION
Objective. The purpose of this study was to examine the association between tobacco marketing efforts and daily cigarette smoking by adolescents. Design. This was a longitudinal study of uptake of smoking on a daily basis with smoking data from the Monitoring the Future project. Diffusion modeling was used to generate expected rates of daily smoking initiation, which were compared with actual rates. Study data were from a national survey, administered annually from 1978 through 1995. Between 4,416 and 6,099 high school seniors participated per year, for a total of 94,652. The main outcome measure was a deviation score based on expected rates from diffusion modeling vs actual rates of initiation of daily use of cigarettes by ninth graders. Annual data on cigarette marketing expenditures were reported by the Federal Trade Commission. Results. The deviation scores of expected vs actual rates of smoking initiation for ninth graders were correlated with annual changes in marketing expenditures. The correlation between sales promotion expenditures and the deviation score in daily smoking initiation was large (r 5 0.769) and statistically significant (P 5 0.009) in the 1983–1992 period. Correlations between sales promotion and smoking initiation were not statistically significant in 1978–1982. Correlations between advertising expenditures and smoking initiation were not significant in either period. Conclusions. In years of high promotional expenditures, the rate of daily smoking initiation among ninth graders was higher than expected from diffusion model predictions. Large promotional pushes by cigarette marketers in the 1980s and 1990s appear to be linked with increased levels of daily smoking initiation among ninth graders. q1999 American Health Foundation and Academic Press Key Words: cigarette smoking; adolescent smoking; tobacco marketing; sales promotion; diffusion.
1
Fax: (419) 372-8062. E-mail:[email protected].
The uptake of cigarette smoking among adolescents follows a regular pattern that is well known to health researchers. In particular, first trial of cigarettes rises throughout elementary school, peaks around the 6th grade, and declines annually thereafter [1,2]. The incidence of smoking on a daily basis lags first trial, usually peaking in the 9th grade. This basic pattern has been observed for successive cohorts for many years; that is, it represents a highly stable phenomenon of the teenage years that recurs for each class. Data on grade in which initiation of daily smoking occurs show a marked peak in 9th grade. Ninth grade rates of daily smoking initiation are, on average, 40% higher than 8th-grade and 20% higher than 10th-grade levels, which are in turn higher than 7th- or 11th-grade levels. In the data examined below, 9th grade was the peak year for 15 of the 16 cohorts examined, while the 9th-grade rate in the remaining cohort was near to the peak figure (6.4 vs 6.9% of the class). This pattern reflects a growing propensity to regularize smoking as the cohort progresses through the 9th grade, followed by reduced propensity after the ninth grade. After 9th grade the attractiveness of smoking appears to progressively decline, as the resolve of nonsmokers to remain nonsmokers strengthens. In other words, 9th-grade represents the point of maximum hazard for adolescents becoming daily smokers. A nonsmoker in 9th grade who reaches the 10th grade in that condition is at decreasing risk of becoming a smoker thereafter. Because of this pattern, the 9th-grade population contains a larger proportion than other grades of individuals who are in an intermediate state of inclination with respect to smoking. That is, the 9th grade has a high proportion of individuals lacking either a strong predisposition to start smoking or a strong commitment to avoid smoking. Compared with others having stronger inclinations, these students are expected to be more sensitive to outside influence in their choice of smoking or nonsmoking. The present research examines 9th
243
0091-7435/99 $30.00 Copyright q 1999 by American Health Foundation and Academic Press All rights of reproduction in any form reserved.
244
WILLIAM H. REDMOND
graders precisely because influences on smoking behavior are likely to be more readily detectable. Influences impinging on these “fence sitters” may come from a variety of sources, including friends, entertainers, parents, teachers, or marketers. The present research focuses on marketing influences, with particular attention to sales promotion expenditures. SALES PROMOTION
The influence of tobacco-marketing activities on adolescents has been the subject of discussion and debate for a considerable time. In both public policy forums and the tobacco control literature, the focus has been mainly on the role of advertising. However, a growing body of research shows that a substantial number of adolescents are affected by tobacco promotions [3–5]. Concerns about the effects of promotions on youths are sufficiently strong that the Food and Drug Administration’s proposed rules regulating tobacco marketing have the effect of prohibiting or restricting many of the promotional tactics in current use by tobacco companies, in addition to advertising restrictions [6]. As is the case with many consumer products, advertising and promotion are used in tandem in cigarette marketing [1,3]. Although advertising and promotion have the same ultimate goal of increasing sales, their roles are distinct [7,8]. Advertising is meant to influence perceptual factors such as brand awareness, image, and positioning. That is, the purpose of advertising is to condition the consumer’s perceptual structure in a way that is both positive and enduring. Sales promotion, on the other hand, is more closely linked to actions than perceptions. Specifically it is meant to generate trial of the product or influence selection of the brand over other brands. Sales promotion is aimed at stimulating behaviors rather than conditioning perceptions and is expected to have more of a near-term impact than a long-run effect. Sales promotion tactics include coupons, free samples, premiums, in-store displays, and a variety of other methods and are notable for visibility and for presence at the point of purchase. Unlike the marketers of most other consumer products, however, tobacco companies do not advertise on radio and television, so that their reliance on promotional activities is comparatively greater. In fact, expenditures for sales promotion of cigarettes have grown to be substantially higher than advertising expenditures in the United States. Figure 1 illustrates trends in advertising and sales promotion after 1970, which was the last year of radio and television advertising for cigarettes (data are as reported to the Federal Trade Commission, unadjusted for inflation). Because of their heavy reliance on sales promotion,
cigarette marketers employ a very wide range of promotional tactics [2]. These include, in descending order of expenditures, coupons and special offers, promotional allowances to distributors, speciality items (e.g., logo merchandise), point-of-purchase displays, public entertainment (e.g., sports event sponsorship), samples, direct mailing, and audiovisuals. In terms of consumer reaction, the categories of coupons, special offers, and samples have the effect of lowering the effective price of smoking, while promotional allowances and point-ofpurchase displays have the effects of wider availability, favorable shelf location, and enhanced in-store visibility. Sponsorships serve to keep the brand name visible and linked to desirable activities; specialty items and direct mail also keep the brand visible and frequently link the product to peers and reference groups. Cigarette marketing campaigns may or may not be intentionally aimed at adolescents, but it is clear that such efforts are seen by youths and are attractive to them: survey evidence shows high rates of participation in tobacco promotions by adolescents. Over one-third of minors reported some level of participation, with about 1 in 10 reporting possession of one or more promotional items [4]. Willingness to use a promotion item exceeds 25% among junior high and high school-aged teenagers [3]. Awareness of and participation in tobacco promotional campaigns has been found to be significantly associated with susceptibility to tobacco use, whether examined alone or in conjunction with advertising effects [9,10]. In sum, the evidence shows that youths are attracted by tobacco promotions and youths with a higher degree of involvement are more susceptible to smoking. UPTAKE OF SMOKING
The process of smoking initiation among adolescents has been adequately examined elsewhere and is discussed here only in a limited sense, specifically to establish a methodological linkage to diffusion. A number of researchers have observed that the spread of smoking among a cohort of adolescents bears similarities to the spreading acceptance of a new product in a market. That is, the spread of smoking is a diffusion process [11,12]. While cigarettes have been on the market for a considerable time, they are a new experience for each new group of adolescents, and the adolescents react to these as such. Similarities between the uptake of smoking and the spreading acceptance of new products are multiple. Prospective smokers recognize both advantages and drawbacks of smoking and a portion choose to adopt because of the perceived advantages while others are deterred by a focus on the drawbacks. Prospective smokers are influenced by multiple sources, including the actions and opinions of peers, role models, and marketers [13].
EFFECTS OF TOBACCO PROMOTION ON NINTH GRADERS
245
FIG. 1. Advertising and sales promotion expenditures: 1971–1994. Source: Federal Trade Commission.
Also, the spread of smoking among a cohort of adolescents has contagion-like properties, a hallmark of the diffusion process [14]. A small proportion of students begin smoking on a regular basis in elementary school; more subsequently follow the example of these early smokers and, in turn, influence the uptake of smoking by still other students. As noted above, the diffusion of smoking among adolescents has been found to be a highly predictable phenomenon. Experimenting with cigarettes starts at an earlier age than frequent smoking, with the peak time of first trial at sixth or seventh grade [15]. Recently, about 2 or 3% of the class have become daily smokers by the sixth grade, with an increasing rate through the ninth grade, then a decreasing rate through high school and beyond. The main difference for cohorts prior to the mid-1980s is that the sixth-grade rate was closer to 5% and a higher overall percentage became smokers. However, the basic pattern of diffusion was the same, with peak rates of initiation of daily smoking in the ninth grade. The methodological link consists in the use of diffusion modeling techniques to quantitatively characterize the spread of smoking in each of several cohorts of teenagers. Diffusion models of the kind used here have become standard techniques for analyzing and forecasting the diffusion of new products [16]. The technique has also been applied to a number of nonproduct situations, including the spread of smoking cessation among adults [17]. For the present research, diffusion parameter estimates are used to establish a baseline against which actual smoking initiation is compared. In other words, diffusion modeling furnishes a set of expected values for each cohort with which to compare actual
smoking data. Because each cohort can and does have slightly different rates of diffusion, the modeling approach employed here has the advantage of being a more sensitive indicator than alternative methods such as cross-cohort averages or extrapolations. DESIGN, DATA, AND METHODS
Design The study is designed to assess the impact of marketing activities on daily smoking initiation of ninth graders by making two types of comparisons. First is the influence of advertising compared with sales promotion. Second is a comparison of two time periods in which the basic characteristics of marketing strategy changed. The design is responsive to calls for research on smoking that utilizes large samples of adolescent smokers and nonsmokers, that analyzes longitudinal data, and that employs advanced statistical models [1,18]. Eighteen cohorts were separated into two groups based on their exposure to different regimes of marketing strategy. Eight of these are from the period in which sales promotion was less than advertising; these were ninth graders in 1978–1982. Ten cohorts are from the period in which sales promotion exceeded advertising; these were ninth graders in 1983–1992. For each of the 18 cohorts, a separate diffusion model was estimated from survey returns of daily smoking initiation in grades 6 through 12. The diffusion model established a “normal” or expected rate of smoking initiation for the ninth grade. The quantity of interest is the difference between the expected rate and the actual rate of initiation in the ninth grade. Thus the main criterion variable
246
WILLIAM H. REDMOND
is a deviation score representing the extent to which daily smoking initiation was higher or lower than the level that would be expected if the diffusion process was operating normally. Figure 2 shows actual and expected values for one of the cohorts. For the cohorts in each group, the degree of association between advertising changes, promotional changes, and deviation score was assessed by correlation. Data Data for daily smoking initiation are from the University of Michigan’s Monitoring the Future project, which examines a wide variety of substance use behaviors and related attitudes among high school seniors [e.g., 19]. The question used in the present research asks respondents to report the grade in which they first began to smoke cigarettes on a daily basis. This question is item 5570 and appears on two of the five survey forms (in the 1992 returns, these are questions B100A of Form 1 and E10A of Form 3). Sample sizes responding to this question varied by year, from 4,416 to 6,099, for a total of 94,652 in the 18 cohorts. Daily use of cigarettes is associated with the daily nicotine addiction cycle [2]. That is, this measure differs from a first-use type of measure in that it reflects regular use and dependence rather than occasional use or experimenting with cigarettes. Data on cigarette marketing expenditures are reported by the Federal Trade Commission [1,20]. Advertising expenditures include expenditures for newspaper, magazine, billboard, and public transit placements.
Promotional expenses include the categories described above. The data used for this research were year-toyear changes in expenditures, expressed as a percentage. Data reported for 1991 and 1992 were adjusted to correct for the effects of a change in the pricing strategy of cigarette companies [21]. This situation involved the reaction of “name brand” cigarette marketers to the market share erosion caused by generics and private label cigarettes [22]. The price strategy of name brands was to increase their coupon activity to the degree that it represented a broad, but temporary, price reduction. Finding this less than fully satisfactory, the name brand marketers switched to a permanent price cut in 1993, at which time overall industry coupon expenditures fell back to their previous levels [23,24]. The data adjustment was to fix coupon expenditures for 1991 and 1992 at levels that prevailed in 1990 and 1994; other categories were not affected. That is, the purpose was to remove a transient pricing strategy from the promotional classification. Model The diffusion framework used to evaluate the data is the Bass [25] diffusion model, which is the standard methodology for assessing diffusion in marketing [14,16]. This specific model is well known in the literature on new product diffusion; however, the basic model form is closely related to contagion modeling in epidemiology. In this model, the number of new adopters in period t, (St), is a function of the eventual number of adopters in the cohort (m), the number of individuals
FIG. 2. Actual vs expected daily smoking initiation for the high school class of 1988.
EFFECTS OF TOBACCO PROMOTION ON NINTH GRADERS
who have already adopted (Yt21), and the probability that a nonadopter adopts in period t (Pt),
247
prevalent in marketing campaigns, the results of the present study may understate the effects of advertising and/or promotion on teenage smoking.
St 5 (m 2 Yt21)Pt , RESULTS
where Pt 5 p 1 q [Yt21 /m]. Here, p is the innovation parameter and q is the imitation parameter. The innovation parameter represents the earliest starters, while q represents the basic diffusion rate, that is, the rate of spread among the main group of potential adopters. Parameters of the model were estimated using the maximum likelihood procedure outlined by Takada and Jain [26] on SAS NLIN software. Limitations Interpretation of the results is made within the following limitations. Self-reported measures of smoking have been found to be accurate when checked against chemical evidence [27]. However, the data on smoking initiation were obtained in a recall format and may therefore be subject to inaccurate recollection as to the grade of initiation [15]. Also, some respondents may have started daily smoking in the interim between one grade and the next, and made an arbitrary assignment as to grade. For these kinds of errors, the magnitude is uncertain but is thought not to be large, and there is no indication of a systematic or directional bias. A second limitation is that the smoking initiation data have a directional bias; however, the magnitude of this error is well understood. This is a result of surveying 12th graders thereby excluding high school dropouts, who are known to exhibit relatively higher smoking rates and earlier initiation times than high school graduates. Had dropouts been included in the survey, initiation rates for all grades except the 12th would be higher, with the 9th-grade rate 4.0–4.5% higher [15]. Such an inclusion would result in modest increases in model parameters p and m. Absent evidence that dropouts react in a systematically different way to marketing appeals, there is no basis upon which to anticipate substantively altered correlation results. Last, because the advertising and sales promotion data are based on expenditures, these measures reflect the likelihood of exposure. In other words, the measures do not account for qualitative aspects of exposure and therefore are not sensitive to the presence of marketing campaigns that are either especially appealing to youths or off-putting to youths. Previous studies indicate that certain types of content are associated with higher brand awareness and market share among young smokers [28]. To the extent that such content is
Diffusion Modeling Model diagnostics indicate a good fit of the smoking data with the basic diffusion process of increasing rates of initiation followed by decreasing rates. Goodness of fit, as measured by R 2, ranged between 0.92 and 0.99 for the 18 models, which is slightly above average values for diffusion models based on new consumer products [29]. Across time, parameter estimates reflected broad trends in teenage smoking, with an overall decline in smoking rates through the 1980s and the start of an increase in the early 1990s. Among the 18 cohort models, estimates for the innovation parameter ( p) ranged between 0.01 and 0.05; estimates for long-run daily smoking potential (m) ranged between 28.9 and 50.4. The estimates for longrun potential can be interpreted as percentages of the cohort: in the earliest cohort, slightly over 50% were estimated to become daily smokers at some point in their lives. Estimates for the imitation parameter (q) below are rates and are not interpretable as percentages of the cohort. Estimates for the imitation parameter (q) ranged between 0.612 and 0.771. The imitation parameter (q) is frequently a main focus of interest in diffusion studies because it reflects the speed with which a new product spreads within a group. For comparative purposes, the q parameter for new products in the United States averages about 0.30 [30]. In contrast, the q parameter for smoking cessation among adults averages about 0.19 [17]. Compared with the diffusion of new consumer products generally, the uptake of smoking among adolescents is a relatively rapid process, while quitting smoking is a relatively slow process. The criterion variable is a deviation score of expected vs actual smoking. This is computed by subtracting the model prediction of percentage initiating daily smoking in the ninth grade from the actual percentage initiating daily smoking in the ninth grade. This quantity was positive for all years, reflecting the sharp peak that characterizes uptake of daily smoking in the ninth grade and ranged between 10.25% and 11.48%. These data are shown in Fig. 3. The deviation scores are higher on average in the 1983–1992 period than in the 1975– 1982 period (10.66% vs 1 0.57%) and the standard deviation is also larger (0.31 vs 0.22); however, differences between the two periods are not statistically significant at P 5 0.10 or better.
248
WILLIAM H. REDMOND
FIG. 3. Ninth-grade smoking deviation scores by year.
Correlations Associations between marketing activities and daily smoking were estimated by standard Pearson productmoment correlation techniques. Results are shown in Table 1. For the 1975–1982 period, the correlation coefficient between advertising expenditures and the smoking deviation score is larger than the coefficient between sales promotion expenditures and the smoking deviation score, although neither is significant at P 5 0.10 or better. For the 1983–1992 period, the correlation coefficient between sales promotion expenditures and the smoking deviation score is larger than the coefficient between advertising and smoking and is statistically significant. This correlation coefficient is 0.769 and is significant at P 5 0.009. The correlation coefficient between smoking and advertising is not statistically significant in this time period. The relative strength of association between advertising and the smoking deviation score is reversed in the two time periods, which moved in the opposite direction of the association between promotion and the smoking deviation score. For the period in which sales promotion expenditures were lower than advertising (1975–1982), the association between promotional expenditures and the smoking deviation score is relatively weak r 5 0.287; P 5 0.49). TABLE 1 Correlation Coefficients Relating Deviation in Daily Smoking Initiation to Sales Promotion and to Advertising Expenses Correlation between smoking deviation score and
9th graders in 1975–1982
9th graders in 1983–1992
Sales promotion expenditures
0.287 (P 5 0.490) 0.534 (P 5 0.172)
0.769 (P 5 0.009) 0.246 (P 5 0.493)
Advertising expenditures
For the period in which sales promotion expenditures exceeded advertising (1983–1992), the association between sales promotion expenditures and the smoking deviation score is positive, substantial, and statistically significant. The magnitude of the correlation coefficient indicates that year-to-year changes in sales promotion expenditures account for about 54% of the variance in the smoking deviation score. CONCLUSION
Prior research has established that cigarette sales promotions are attractive to adolescents and that receptivity to promotional items by adolescents is associated with progression toward smoking behavior [3,5]. The present research establishes a linkage between cigarette promotions and escalated levels of smoking behavior among adolescents. Significantly greater numbers of ninth graders initiate daily smoking in years of high promotional expenditures than would otherwise be expected. The shift by tobacco marketers from an emphasis on advertising to an emphasis on sales promotion appears to have altered the nature of ninth graders’ response to marketing efforts. Since the early 1980s, year-toyear changes in sales promotional expenditures for cigarettes have been associated with deviations in the percentage of ninth graders who start smoking on a daily basis. Previously, such linkages were not statistically significant. The role of sales promotion as a behavioral trigger is manifest in the data only after the point in time at which promotional expenditures came to dominate the marketing strategy of tobacco companies. Whether or not aimed at adolescents, promotional activities are linked to elevated levels of daily smoking in that age group. The pronounced upward trend in sales promotion expenditures by cigarette companies is accompanied by several instances of large promotional “pushes” in which year-to-year increases of $400 million
EFFECTS OF TOBACCO PROMOTION ON NINTH GRADERS
or more are observed. The magnitude of effects is illustrated by figures from the year 1985. This year was the largest percentage increase of the 1980s: promotional expenditures in 1985 rose by $479 million over the 1984 level of $1.0 billion. In 1985 the ninth-grade class had an excess in daily smoking initiation of 0.86% after correcting for the ninth-grade peak factor (the average ninth-grade deviation is 10.62% while the deviation in 1985 is 11.48%). Assuming a cohort size of approximately 3.4 million students, this excess initiation rate translates into roughly 29,000 daily smokers above the level that would otherwise be expected. Further research will be required to ascertain whether sales promotions have a differential impact on racial, ethnic, or gender subgroups or whether the various forms of sales promotion have a differential effect on adolescent smoking. Design Considerations By design, this research is sensitive to deviations in rates of daily smoking initiation, rather than in base rates of daily smoking initiation. It is therefore possible that sales promotion also plays a more fundamental role in daily smoking uptake, but one which is undetectable in the present design. That is, it may be that promotional activities contribute to an overall predisposition to smoke and have, as a result, elevated the “expected” rate of daily smoking above rates that would be observed in the absence of large promotional campaigns. As noted above, the design precludes an evaluation of the qualitative aspects of promotional activities, as well as nonmarketing sources of influence on smoking behaviors. Lack of significance of the advertising variable does not necessarily imply an absence of causative links between advertising and daily smoking initiation. Advertising is expected mainly to have a long-term, cumulative effect in shaping attitudes and images. The present results indicate that year-to-year changes in advertising expenditure are not significantly associated with deviations in daily smoking initiation in that same year. The literature on adolescent smoking typically positions advertising as being among the determinants of predisposition to smoke, that is, one factor in a longterm accumulated inclination [10]. However, research on alcohol advertising indicates that geographic or temporal aggregation of data tend to obscure the linkages [31]. It is thus possible that a disaggregated approach may prove more insightful in cigarette advertising. ACKNOWLEDGMENT The author thanks Floris Wood and Darryl Stone for valued assistance in data retrieval.
249
REFERENCES 1. U.S. Department of Health and Human Services. Preventing tobacco use among young people: a report of the Surgeon General. Atlanta: U.S. Department of Health and Human Services, 1994. 2. Lynch B, Bonnie R, editors. Growing up tobacco free: preventing nicotine addiction in children and youths. Washington: National Academy Press, 1994. 3. Gilpin E, Pierce J, Rosbrook B. Are adolescents receptive to current sales promotion practices of the tobacco industry? Prev Med 1997;26:14–21. 4. Coeytaux R, Altman D, Slade J. Tobacco promotion in the hands of youth. Tobacco Control 1995;4:253–7. 5. Pierce J, Choi W, Gilpin EA, Farkas A, Berry L. Tobacco industry promotion of cigarettes and adolescent smoking. JAMA 1998;297:511–5. 6. Gostin L, Arno P, Brandt A. FDA regulations of tobacco advertising and youth smoking: historical, social and constitutional perspectives. JAMA 1997;277:410–8. 7. Kotler P. Marketing management: analysis, planning, implementation and control. 9th ed. Upper Saddle River (NJ): Prentice Hall, 1997. 8. Farris P, Quelch J. Advertising and promotion management: a manager’s guide to theory and practice. Radnor (PA): Chilton, 1983. 9. Altman D, Levine D, Coeytaux R, Slade J, Jaffe R. Tobacco promotion and susceptibility to tobacco use among adolescents aged 12 through 17 years in a nationally representative sample. Am J Public Health 1996;86:1590–3. 10. Evans N, Farkas A, Gilpin E, Berry C, Pierce J. Influence of tobacco marketing and exposure to smokers on adolescent susceptibility to smoking. J Nat Cancer Inst 1995;87:1538–45. 11. Pierce J. Progress and problems in international public health efforts to reduce tobacco usage. In: Omenn G, editor. Annual review of public health, Vol. 12. Palo Alto (CA): Annual Reviews, 1991:383–400. 12. Novak T, Bass F, Kamakira W, Dent C. Diffusion models for smoking onset and cessation: a segment-level historical analysis. In: McAlister L, Rothschild M, editors. Advances in consumer research, Vol. XX. Provo (UT): Association for Consumer Research, 1993. 13. Elders MJ, Perry C, Eriksen M, Giovino G. The report of the Surgeon General: preventing tobacco use among young people. Am J Public Health 1994;84:543–7. 14. Rogers E. Diffusion of innovations. 4th ed. New York: The Free Press, 1995. 15. Johnston L, O’Malley P, Bachman J. National survey results on drug use. Washington: U.S. Department of Health and Human Services, 1996. 16. Mahajan V, Mueller E, Bass F. New product diffusion models in marketing: a review and directions for research. J Market 1990;54:1–26. 17. Redmond W. Product disadoption: quitting smoking as a diffusion process. J Public Policy Market 1996;15:87–97. 18. Pollay R. Hacks, flacks, and counter-attacks: cigarette advertising, sponsored research, and controversies. J Social Issues 1997;53:53–74. 19. Bachman J, Johnston L, O’Malley P. Monitoring the future: questionnaire responses from the nation’s high school seniors. Ann Arbor (MI): Univ. of Michigan, 1993. 20. Federal Trade Commission. Report to Congress for 1990: pursuant to the Federal Cigarette Labeling and Advertising Act. Washington: Federal Trade Commission, 1992.
250
WILLIAM H. REDMOND
21. When smoke got in their eyes. Economist 1993;327:65–6. 22. Low-price smokes fire up promo flurry. Advertising Age 1990;61:33–4. 23. Phillip Morris making lower prices for Marlboro cigarettes permanent. New York Times 1993 Jul 21. 24. Add RJR to list of cig price cuts. Advertising Age 1993;64:3/46. 25. Bass F. A new product growth model for consumer durables. Manage Sci 1969;16:215–27. 26. Takada H, Jain D. Cross-national analysis of diffusion of consumer durable goods in Pacific rim countries. J Market 1991;55:48–54. 27. Patrick D, Choadle A, Thompson D, Diehr P, Koepsell T, Kinne
S. The validity of self-reported smoking: a review and metaanalysis. Am J Public Health 1994;84:1086–93. 28. Pollay R, Lavack A. The targeting of youths by cigarette marketers: archival evidence on trial. In: McAlister L, Rothschild M, editors. Advances in consumer research, Vol. XX. Provo (UT): Association for Consumer Research, 1993. 29. Wind Y, Mahajan V, Cardozo R, New product forecasting. Lexington (MA): Lexington, 1981. 30. Sultan F, Farley J, Lehmann D. A meta-analysis of applications of diffusion models. J Market Res 1990;27:70–7. 31. Saffer H. Alcohol advertising and motor vehicle fatalities. Rev Econ Stat 1997;79:431–42.