ARTICLES

Effect of viewing smoking in movies on adolescent smoking

initiation: a cohort study

Madeline A Dalton, James D Sargent, Michael L Beach, Linda Titus-Ernstoff, Jennifer J Gibson, M Bridget Ahrens,
Jennifer J Tickle, Todd F Heatherton

Summary Introduction
Many studies have linked tobacco marketing with an
Background Exposure to smoking in movies has been linked increased risk of smoking uptake in adolescents.1–7 For
with adolescent smoking initiation in cross-sectional studies. example, owning tobacco promotional items and being
We undertook a prospective study to ascertain whether able to recall cigarette advertisements can double the odds
exposure to smoking in movies predicts smoking initiation. that an adolescent will become an established smoker.3
Movie images, like commercial advertising, associate
Method We assessed exposure to smoking shown in movies smoking with celebrities and depict it as an attractive
in 3547 adolescents, aged 10–14 years, who reported in a behaviour.8 In popular contemporary movies, smoking is
baseline survey that they had never tried smoking. Exposure frequently associated with characteristics many
to smoking in movies was estimated for individual adolescents find appealing—such as toughness, sexiness,
respondents on the basis of the number of smoking and rebelliousness.9 Endorsement of cigarette brands in
occurrences viewed in unique samples of 50 movies, which movies by actors has also increased substantially over the
were randomly selected from a larger sample pool of popular past decade.10
contemporary movies. We successfully re-contacted Several studies have described how smoking is
2603 (73%) students 13–26 months later for a follow-up portrayed in movies,9,11–16 but only a few have specifically
interview to determine whether they had initiated smoking. assessed whether viewing smoking in movies affects
adolescent smoking behaviour. In an experimental study,
Findings Overall, 10% (n=259) of students initiated smoking Pechmann and Shih17 showed that adolescents were more
during the follow-up period. In the highest quartile of likely to report positive attitudes toward smoking after
exposure to movie smoking, 17% (107) of students had seeing smoking portrayed in movies. Results of two cross-
initiated smoking, compared with only 3% (22) in the lowest sectional studies18,19 indicated that adolescents were more
quartile. After controlling for baseline characteristics, likely to have tried smoking if their favourite movie stars
adolescents in the highest quartile of exposure to movie smoked on screen. In our previous study of adolescents in
smoking were 2·71 (95% CI 1·73–4·25) times more likely to New England, USA, exposure to smoking in movies was
initiate smoking compared with those in the lowest quartile. associated with smoking experimentation, even after
The effect of exposure to movie smoking was stronger in controlling for the effects of other social influences,
adolescents with non-smoking parents than in those whose parenting, and personality characteristics of the child.20
parent smoked. In this cohort, 52·2% (30·0–67·3) of Collectively, these results suggest that movie smoking
smoking initiation can be attributed to exposure to smoking influences adolescent smoking behaviour. However, the
in movies. cross-sectional design of these studies precludes
establishment of a temporal relation. To determine
Interpretation Our results provide strong evidence that whether exposure to movie smoking predicts smoking
viewing smoking in movies promotes smoking initiation initiation in adolescents, we did a longitudinal study of
among adolescents. adolescents in New England, USA, who had never
previously tried smoking.
Published online June 10, 2003
http://image.thelancet.com/extras/03art1353web.pdf Methods
See Commentary Participants
In 1999, we distributed a self-administered written survey
to adolescents (aged 10–14 years) enrolled in grades 5
through 8 at 14 schools in Vermont and New Hampshire,
USA. The purpose of this baseline survey was to assess
exposure to smoking in movies and investigate its
Departments of Paediatrics (M A Dalton PhD, Prof J D Sargent MD), association with lifetime smoking experience. Details of
Anaesthesia (M L Beach PhD), and Community and Family Medicine the methods for the survey have been published
(M L Beach, L Titus-Ernstoff PhD, J J Gibson MS), Norris Cotton Cancer previously.20
Center, Dartmouth Medical School, One Medical Center Drive, Through the baseline survey, we identified
Lebanon; Department of Health, Social, and Economic Research, 3547 adolescents who had never tried smoking cigarettes
RTI International, Research Triangle Park, NC, USA and were thus eligible for a follow-up 13–26 months later
(M B Ahrens MPH); and Department of Psychological and Brain to assess risk factors for smoking initiation. The follow-up
Sciences, Dartmouth College, Hanover, NH, USA (J J Tickle PhD, telephone interviews, accomplished for 2603 (73%)
Prof T F Heatherton PhD) eligible baseline participants, were done by trained
Correspondence to: Dr Madeline A Dalton, Department of interviewers using a computer-assisted telephone
Paediatrics, Dartmouth Medical School, One Medical Center Drive, interview system. To protect confidentiality, students
Lebanon, NH 03756, USA indicated their answers by pressing numbers on the
(e-mail: [email protected]) telephone. We used a PC Telecom digit grabber

THE LANCET • Published online June 10, 2003 • http://image.thelancet.com/extras/03art1353web.pdf 1

For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES

(Metrotel, Milpitas, CA) so that every time a student (or primary caregiver if they did not have a mother).
pressed a number, the answer was automatically entered Summary measures were created by adding their
into the database. The protocol for this study was responses to each of the individual items, so that higher
approved by the Dartmouth committee for the protection scores signify more of each characteristic. We then divided
of human subjects. the scores into quartiles.

Procedures Statistical analysis

We assessed lifetime smoking experience at baseline and Preliminary analyses consisted of descriptive frequencies,
follow-up by asking “How many cigarettes have you 2 tests to compare differences in proportions, and t tests
smoked in your life?”, to which respondents could answer to compare mean differences by group. We used
“none”, “just a few puffs”, “one to 19 cigarettes”, “20 to generalised linear models28 to assess smoking initiation as
100 cigarettes”, or “more than 100 cigarettes”. Only a function of both movie exposure and baseline covariates.
students who answered “none” at baseline were eligible We used a log link, rather than a logistic regression, so
for follow-up. Students who reported any cigarette that relative risks could be estimated directly. An
smoking (just a few puffs, one to 100 cigarettes, more overdispersion variable was used to account for possible
than 100 cigarettes) on the follow-up survey were clustering by schools. Exposure to movie smoking was
classified as having initiated smoking during the follow-up treated as a categorical variable. The dependent variable
period. was whether the respondent had initiated smoking during
Adolescents’ exposure to smoking in movies was the follow-up period. We did multivariate analyses with
assessed at baseline by asking each student to indicate both minimally adjusted (age, sex, and school) and fully
which films he or she had seen from a unique list of adjusted models. The fully adjusted models included all
50 movies. A list of 50 movies was randomly selected for terms for child characteristics, social influences, and
each individual survey from a sample of 601 popular parenting characteristics as described above, as well as the
contemporary movies released between 1988 and 1999. time elapsed between the baseline and follow-up surveys.
The 601 movies included the top 25 box-office hits every We assessed model fit and interaction terms with changes
year from 1988 to 1995 (n=200); the top 100 box-office in deviances and standard diagnostic plots. Results were
hits per year from 1996 to 1998 (300); the top 50 box- judged significant if p<0·05, in a two-sided test.
office hits from the first half of 1999; and 51 additional Simulation methods, similar to those used by Connors
movies selected because they featured stars popular and colleagues,29 were used to test whether an
among adolescents. We stratified the random selection of unmeasured confounder could falsely implicate movie
movies so that each list of 50 had the same distribution of exposure. Attributable risk was estimated by the
ratings as the larger sample of top box-office hits: 45% R probability of initiating smoking for each adolescent,
(restricted, younger than 17 years requires accompanying assuming varying degrees of movie exposure and holding
parent or adult guardian), 31% PG-13 (parents strongly measured covariates constant.
cautioned, some material might be inappropriate for
children younger than 13 years), 20% PG (parental Role of the funding source
guidance suggested, some material might not be suited for The sponsor of the study had no role in study design, data
children), 4% G (general audiences, all ages admitted). collection, data analysis, data interpretation, or writing of
On average, every movie title was included in the report.
470 questionnaires. Trained coders counted the number
of occurrences of smoking in each movie using methods Results
previously described.9 We calculated exposure to movie Our final sample of 2603 adolescents was mainly white
smoking for each respondent by summing the number of (94%, n=2392), as was the underlying population (96%);
smoking occurrences for each movie the respondent had equally distributed by sex; with a mean age at baseline of
seen. We adjusted for possible variation in the movie lists 12 years (SD 1·1). Participants who were followed up
by expressing individual exposure to movie smoking as a were much the same as non-participants in age, sex,
proportion of the total number of possible smoking grade, and exposure to movie smoking, but non-
occurrences each student could have seen on the basis of participants were more likely than participants to have
the movies included in their survey. Exposure to movie parents who smoke (41% [383] vs 30% [773],
smoking was classified in quartiles with the following respectively) and slightly more likely to be susceptible to
cutoffs: 0–531 occurrences for the 1st quartile, 532–960 smoking30 at baseline (27% [257] vs 23% [592]); report
for the 2nd quartile, 961–1664 for the 3rd quartile, and average or below average school performance (25% [237]
1665–5308 for the 4th quartile. vs 19% [484]); have friends who smoke (30% [282] vs
We also measured at baseline, through questions 26% [671]); and have siblings who smoke (14% [134] vs
adapted from previously validated questionnaires, 10% [267]). Reasons for non-participation included
variables that could potentially confound the association refusal to provide contact information at baseline (35%,
between movie exposure and adolescent smoking 326), refusal to participate in the interview at the time of
initiation. These variables included child characteristics follow-up (31%, 288), and lost-to-follow-up (35%, 330).
(sex, age, school, self-reported school performance, On average, students had seen 16 of the 50 movies they
sensation seeking,21,22 rebelliousness,23 and self-esteem24), were asked about, from which they were exposed to an
social influences (parent, sibling, and friend smoking; average of 98·5 (SD 75·1) smoking occurrences. Exposure
receptivity to tobacco promotions4,25), and parenting to movie smoking increased with age and was higher in
characteristics (parent education, two measures of boys than in girls. Girls saw a mean of 14·6 movies (7·4),
authoritative parenting,26 and adolescents’ perception of from which they viewed a mean of 85·1 smoking
parental disapproval of smoking27). Individual items used occurrences (66·4), whereas boys saw a mean of 17·1
to measure student personality and parenting movies (8·2), from which they viewed 113·5 smoking
characteristics have been reported previously.20 Students occurrences (81·2). Exposure to movie smoking was
used a four-point response scale to indicate how well positively associated with sensation seeking (p<0·0001)
specific statements described themselves or their mothers and rebelliousness (p<0·0001), and inversely associated

2 THE LANCET • Published online June 10, 2003 • http://image.thelancet.com/extras/03art1353web.pdf

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 ARTICLES

with school performance and measures of authoritative school showed significant associations between baseline
parenting (p<0·0001)). characteristics, including exposure to movie smoking, and
10% (259) of participants initiated smoking during the smoking initiation (table 1). Relative to the lowest quartile
follow-up period. Most (80%, n=208) of those who of movie smoking exposure, the risk for smoking initiation
initiated smoking reported that they had smoked “just a increased with each successive quartile of exposure
few puffs” of a cigarette. Only 2% (six) of those who (table 1). Although the relative risks were attenuated, the
initiated smoking had smoked more than 100 cigarettes relation between exposure to movie smoking and smoking
during follow-up. Analyses adjusted for age, sex, and initiation remained significant after adjustment for all

Characteristic Total (n=2603) Tried smoking Relative risk* (95% CI)

Sociodemographic
Age
10 to <12 years 809 50 (6%) 1·00
12 to <13 years 804 68 (8%) 1·40 (0·98–2·01)
13 to <15 years 990 141(14%) 2·31 (1·67–3·19)
Sex
Male 1234 119 (10%) 1·00
Female 1369 140 (10%) 1·09 (0·87–1·38)
Social influences
Either parent smokes
No 1830 133 (7%) 1·00
Yes 773 126 (16%) 2·25 (1·77–2·86)
Any friends smoke
No 1932 147 (8%) 1·00
Yes 671 112 (17%) 1·87 (1·46–2·41)
Any siblings smoke
No 2336 210 (9%) 1·00
Yes 267 49 (18%) 1·91 (1·42–2·59)
Receptive to tobacco promotions
No 2161 179 (8%) 1·00
Yes 442 80 (18%) 2·09 (1·62–2·71)
Child characteristics
School performance
Excellent 1113 53 (5%) 1·00
Good 1006 114 (11%) 2·29 (1·67–3·13)
Average/below average 484 92 (19%) 3·65 (2·62–5·09)
Sensation seeking
First quartile 792 40 (5%) 1·00
Second quartile 709 59 (8%) 1·60 (1·09–2·35)
Third quartile 484 55 (11%) 2·21 (1·49–3·27)
Fourth quartile 618 105 (17%) 3·27 (2·28–4·68)
Rebelliousness
First quartile 771 37 (5%) 1·00
Second quartile 549 39 (7%) 1·48 (0·96–2·27)
Third quartile 668 71 (11%) 2·24 (1·53–3·29)
Fourth quartile 615 112 (18%) 4·10 (2·84–5·91)
Self-esteem
First quartile 676 100 (15%) 1·00
Second quartile 747 68 (9%) 0·64 (0·48–0·86)
Third quartile 760 71 (9%) 0·68 (0·51–0·92)
Fourth quartile 420 20 (5%) 0·35 (0·22–0·56)
Parent characteristics
Maternal demandingness
First quartile 617 68 (11%) 1·00
Second quartile 666 71 (11%) 0·97 (0·70–1·33)
Third quartile 755 74 (10%) 0·86 (0·63–1·18)
Fourth quartile 565 46 (8%) 0·72 (0·50–1·04)
Maternal responsiveness
First quartile 526 78 (15%) 1·00
Second quartile 571 60 (11%) 0·76 (0·55–1·05)
Third quartile 679 63 (9%) 0·69 (0·50–0·94)
Fourth quartile 827 58 (7%) 0·55 (0·39–0·76)
Parent education
Both completed high school 2223 206 (9%) 1·00
Neither or one completed high school 380 53 (14%) 1·55 (1·15–2·08)
Parental disapproval of smoking
Both disapprove 2157 197 (9%) 1·00
Neither or one disapproves 446 62 (14%) 1·53 (1·16–2·01)
Movie smoking exposure†
First quartile 651 22 (3%) 1·00
Second quartile 651 56 (9%) 2·39 (1·49–3·83)
Third quartile 651 74 (11%) 2·99 (1·89–4·72)
Fourth quartile 650 107 (16%) 4·31 (2·76–6·75)
*Relative risk for age at baseline is adjusted for sex and school. Relative risk for sex is adjusted for age and school. All other relative risks are adjusted for age at
baseline, sex, and school. †First quartile, 0–531 occurrences of smoking; second quartile, 532–960 occurrences; third quartile, 961–1664 occurrences; and fourth
quartile 1665–5308 occurrences.
Table 1: Predictors of smoking initiation

THE LANCET • Published online June 10, 2003 • http://image.thelancet.com/extras/03art1353web.pdf 3

For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES

Quartile of movie smoking exposure*

1 2 3 4
All participants 1·00 2·02 (1·27–3·20) 2·16 (1·38–3·40) 2·71 (1·73–4·25)
Parental smoking
Non-smoker 1·00 2·32 (1·21–4·45) 2·64 (1·39–5·01) 4·08 (2·19–7·61)
Smoker 2·84 (1·28–6·29) 4·77 (2·41–9·44) 4·64 (2·43–8·87) 4·74 (2·49–9·02)
Values are relative risks (95% CI) adjusted for time between surveys and the following baseline characteristics: grade, sex, school, friend smoking, sibling smoking,
parent smoking, receptivity to tobacco promotions, school performance, sensation-seeking propensity, rebelliousness, self esteem, parent education, authoritative
parenting, and perception of parental disapproval of smoking. *First quartile, 0–531 occurrences of smoking; second quartile, 532–960 occurrences; third quartile,
961–1664 occurrences; and fourth quartile 1665–5308 occurrences.
Table 2: Effect of movie smoking exposure on smoking initiation in all participants, and the interaction between movie smoking
exposure and parental smoking in relation to smoking initiation

baseline covariates. Compared with the lowest exposure visual media exposure with high risk behaviour in
level, adolescents in the second, third and fourth quartiles adolescents.31
were two to three times more likely to initiate smoking The data suggest that children with non-smoking
during follow-up (table 2). parents are especially susceptible to the effect of movie
We assessed potential interactions between exposure to smoking exposure. Children with parents who smoke
movie smoking and age, sex, and social influences (friend, might have a more realistic view of smoking, so they are
sibling, and parent smoking) on smoking initiation and less likely to be influenced by the glamorous portrayal of
identified a significant interaction between exposure and smoking in movies. However, an equally plausible
parental smoking behaviour (p=0·003). In adolescents explanation is that children with parents who smoke are
with non-smoking parents, the risk of smoking initiation already at a higher risk for smoking initiation, so their risk
increased substantially with greater exposure to movie is less likely to be raised by other social influences. Further
smoking. Those with smoking parents had an overall research is needed to understand this interaction fully.
higher risk of smoking initiation, but were less influenced Although it is not feasible to completely measure an
by exposure to movie smoking than those whose parents adolescents’ total lifetime exposure to smoking in movies,
did not smoke (table 2). every survey in our study contained 50 randomly selected
Even after controlling for all other covariates, 52·2% movies from a larger sample of 601 films, stratified by rating.
(95% CI 30·0–67·3) of smoking initiation in this cohort Thus, our assessment is an unbiased estimate of adolescents’
can be attributed to exposure to smoking in movies. If the exposure to smoking in popular, contemporary movies.
observed association with smoking initiation is assumed to Unlike most measures of exposure to tobacco marketing, this
be causal, reducing movie smoking exposure in this study assessment reflects actual exposure rather than adolescents’
to the lowest quartile would have reduced the proportion attention, attitudes or predispositions to smoking. However,
who initiated smoking during follow-up from 10·0% to because almost all R-rated movies contain smoking,9 we
4·8%. Reducing movie exposure for all children by just could not separate the effects of an R-rating and smoking
one quartile (eg, moving a child from the fourth to the content. Consequently, we cannot exclude the possibility
third quartile) would correspond to an attributable risk that some other aspect of R-rated movies influences smoking
reduction of 21·4% (12·0–29·8), which would have initiation. However, more than 40 years of research shows
reduced the proportion who initiated smoking in this that observers imitate specific behaviours they see
study from 10·0% to 7·8%. modelled.32,33 Thus, our inference that adolescents imitate
Our simulation studies indicate it is unlikely that an smoking behaviour seen in movies seems reasonable. The
unmeasured covariate was responsible for the association generalisability of our findings might be restricted because
between exposure to movie smoking and smoking initiation. our sample included a mainly white, rural population.
To raise the relative risk to the magnitude we recorded, a The effect of exposure to movie smoking is important,
potential confounder would need to be associated with both both because the effect on smoking initiation is moderately
movie exposure (with a minimum correlation of 0·2) and strong and because the exposure is almost universal. Based
smoking initiation (minimum relative risk of 1·2) and be on the lists of 50 randomly selected movies, only five
independent of all other covariates we measured. An (0·2%) participants were unexposed to movie smoking. If
unmeasured independent covariate would have to have the link between exposure to smoking in movies and
p values of less than 0·00001 associated with both movie smoking initiation proves to be causal, our data suggest
exposure and smoking initiation. This is unlikely because that eliminating adolescents’ exposure to movie smoking
any covariate we did not measure would almost certainly be could reduce smoking initiation by half. However, we
associated with at least one of the measured covariates, so recognise that the equation might not be that simple, since
that a substantial proportion of the variability would already many factors affect movie exposure and its effect on
be accounted for. adolescent behaviour. We controlled for as many of these
factors as possible, and our sensitivity analysis suggests that
Discussion an unmeasured variable is unlikely to account for the
Our results suggest that viewing smoking in movies association between exposure to movie smoking and
strongly predicts whether or not adolescents initiate smoking initiation. Because the follow-up period for this
smoking, and the effect increases significantly with study was brief, we could not assess the possibly greater
greater exposure. Adolescents who viewed the most effects of longer term exposure. Consequently, the effect of
smoking in movies were almost three times more likely to reducing exposure to smoking in movies over many years
initiate smoking than those with the least amount of could be larger than that we recorded. Nonetheless, it is
exposure. The magnitude of this association is consistent important to point out that this study links movie smoking
with the results of our cross-sectional study of exposure with smoking initiation, and not all initiators will
adolescents in New England, USA.20 It is also consistent become established smokers. Further research is needed to
with the results of other cross-sectional studies that have assess the effect of exposure to smoking in movies on long-
linked actor smoking with adolescent smoking18,19 and term smoking behaviour.

4 THE LANCET • Published online June 10, 2003 • http://image.thelancet.com/extras/03art1353web.pdf

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 ARTICLES

Contributors 14 McIntosh WD, Bazzini DG, Smith SM, Wayne SM. Who smokes in
M Dalton, J Sargent, M Beach, J Tickle, and T Heatherton designed the Hollywood? Characteristics of smokers in popular films from 1940 to
study, developed the surveys and content analysis, and directed the 1989. Addict Behav 1998; 23: 395–98.
research. L Titus-Ernstoff provided input on the analytical approach and 15 Hazan AR, Lipton HL, Glantz SA. Popular films do not reflect
co-wrote the report. M Ahrens contributed to survey development and current tobacco use. Am J Public Health 1994; 84: 998–1000.
coordinated survey administration. M Beach and J Gibson did the 16 Terre L, Drabmen RS, Speer P. Health-relevant behaviors in the
statistical analysis. All authors contributed to the interpretation of the data media. J Appl Soc Psychol 1991; 21: 1303–19.
and reviewed the final report. 17 Pechmann C, Shih CF. Smoking scenes in movies and antismoking
advertisements before movies: effects on youth. J Mark 1999; 63:
Conflict of interest statement 1–13.
None declared. 18 Distefan JM, Gilpin EA, Sargent JD, Pierce JP. Do movie stars
encourage adolescents to start smoking? Evidence from California.
Acknowledgments Prev Med 1999; 28: 1–11.
This study was funded by the National Cancer Institute (CA-77026). We 19 Tickle JJ, Sargent JD, Dalton MA, Beach ML, Heatherton TF.
thank E Robert Greenberg for his input on the analysis and presentation Favourite movie stars, their tobacco use in contemporary movies and
of the data, Daniel Nassau and Ezra Hays for coding the movies, its association with adolescent smoking. Tob Control 2001; 10: 16–22.
Bill Harper for building the CATI system, Susan Martin for providing 20 Sargent JD, Beach ML, Dalton MA, et al. Effect of seeing tobacco use
administrative support for the study, and Anna Adachi-Mejia and in films on trying smoking among adolescents: cross-sectional study.
Susan Remacle for preparing the report. BMJ 2001; 323: 1394–97. Available from: URL: http://bmj.com/
cgi/content/full/323/7326/1394? (accessed April 11, 2003).
21 Zuckerman M, Bone RN, Neary R, Mangelsdorff D, Brustman B.
What is the sensation seeker? Personality trait and experience
References correlates of the sensation-seeking scales. J Consult Clin Psychol 1972;
1 Pierce JP, Gilpin EA. A historical analysis of tobacco marketing and 39: 308–21.
the uptake of smoking by youth in the United States 1890–1977. 22 Russo MF, Stokes GS, Lahey BB, et al. A sensation seeking scale for
Health Psychol 1995; 14: 500–08. children: further refinement and psychometric development.
2 Choi WS, Ahluwalia JS, Harris KJ, Okuyemi K. Progression to J Psychopathol Behav Assess 1993; 15: 69–85.
established smoking: the influence of tobacco marketing. 23 Pierce JP, Farkas A, Evans N. Tobacco use in California 1992: a focus
Am J Prev Med 2002; 22: 228–33. on preventing uptake in adolescents. Sacramento, CA: California
3 Biener L, Siegel M. Tobacco marketing and adolescent smoking: more Department of Human Services, 1993.
support for a causal inference. Am J Public Health 2000; 90: 407–11. 24 Carvajal SC, Wiatrek DE, Evans RI, Knee CR, Nash SG.
4 Sargent JD, Dalton MA, Beach ML, Bernhardt AM, Heatherton TF, Psychosocial determinants of the onset and escalation of smoking:
Stevens M. Effect of cigarette promotions on smoking uptake among cross-sectional and prospective findings in multiethnic middle school
adolescents. Prev Med 1999; 30: 320–27. samples. J Adolesc Health 2000; 27: 255–65.
5 Schooler C, Feighery E, Flora JA. Seventh graders’ self-reported 25 Pierce JP, Choi WS, Gilpin EA, Farkas AJ, Berry CC. Tobacco
exposure to cigarette marketing and its relationship to their smoking industry promotion of cigarettes and adolescent smoking. JAMA
behavior. Am J Public Health 1996; 86: 1216–21. 1998; 279: 511–15.
6 Altman DG, Levine DW, Coeytaux R, Slade J, Jaffe R. Tobacco 26 Jackson C, Henriksen L, Foshee VA. The Authoritative Parenting
promotion and susceptibility to tobacco use among adolescents aged Index: predicting health risk behaviors among children and
12 through 17 years in a nationally representative sample. adolescents. Health Educ Behav 1998; 25: 319–37.
Am J Public Health 1996; 86: 1590–93. 27 Sargent JD, Dalton MA. Does parental disapproval of smoking
7 Evans N, Farkas A, Gilpin E, Berry C, Pierce JP. Influence of tobacco prevent adolescents from becoming established smokers? Pediatrics
marketing and exposure to smokers on adolescent susceptibility to 2001; 108: 1256–62.
smoking. J Natl Cancer Inst 1995; 87: 1538–45. 28 McCullagh P, Nelder JA. Generalized linear models. London:
8 Basil M. The danger of cigarette “special placements” in film and Chapman and Hall, 1989.
television. Health Commun 1997; 9: 190–98. 29 Connors AF Jr, Speroff T, Dawson NV, et al. The effectiveness of
9 Dalton MA, Tickle JJ, Sargent JD, Beach ML, Ahrens MB, right heart catheterization in the initial care of critically ill patients.
Heatherton TF. The incidence and context of tobacco use in popular SUPPORT Investigators. JAMA 1996; 276: 889–97.
movies from 1988–1997. Prev Med 2002; 34: 516–23. 30 Pierce JP, Choi WS, Gilpin EA, Farkas AJ, Merritt RK, Giovino G.
10 Sargent JD, Tickle JJ, Beach ML, Dalton MA, Ahrens MB, Validation of susceptibility as a predictor of who takes up smoking in
Heatherton TF. Brand appearances in contemporary films and the United States. Health Psychol 1996; 15: 355–61.
contribution to global marketing of cigarettes. Lancet 2001; 357: 29–32. 31 Klein JD, Brown JD, Childers KW, Oliveri J, Porter C, Dykers C.
11 Escamilla G, Cradock AL, Kawachi I. Women and smoking in Adolescents’ risky behavior and mass media use. Pediatrics 1993; 92:
Hollywood movies: a content analysis. Am J Public Health 2000; 90: 24–31.
412–19. 32 Bandura A. Influence of models’ reinforcement contingencies on the
12 Goldstein AO, Sobel RA, Newman GR. Tobacco and alcohol use in acquisition of imitative responses. J Abnorm Psychol 1965; 66: 575–82.
G-rated children’s animated films. JAMA 1999; 281: 1131–36. 33 Bushman BJ, Anderson CA. Media violence and the American public:
13 Everett SA, Schnuth RL, Tribble JL. Tobacco and alcohol use in top- scientific facts versus media misinformation. Am Psychol 2001; 56:
grossing American films. J Community Health 1998; 23: 317–24. 477–89.

THE LANCET • Published online June 10, 2003 • http://image.thelancet.com/extras/03art1353web.pdf 5

For personal use. Only reproduce with permission from The Lancet Publishing Group.