Objective Portrayal of tobacco use in films has been causally linked to youth smoking initiation. However, findings regarding trends in portrayal in US films since 1950 are inconsistent, potentially due to differences in sampling densities, intercoder reliabilities and time periods covered. The present study was designed to overcome these inconsistencies with a common sampling frame and methodology.
Methods A half sample of the 30 top-grossing US films per year from 1950 to 2006 (N=855) was coded in 5-min segments for total tobacco-related content and main character tobacco use. Film tobacco trends were identified using linear regression and compared to national per capita cigarette consumption and historically significant tobacco control events.
Results Tobacco content declined considerably since 1950. Total tobacco-related content peaked around 1961, while the decline in portrayal of main character use was already underway in 1950. Cigarette consumption peaked around 1966 with a trend that closely paralleled total tobacco content and that coincided with major tobacco control events.
Conclusions This study, which had high reliability, dense sampling and covered a long time period, indicates that tobacco content has declined in top-ranked US movies since 1950 with a trend in total tobacco content that closely paralleled the drop in per capita cigarette consumption and the increase in significant tobacco control efforts. Despite the inability to draw causal conclusions, tobacco portrayal in films may serve as barometer of societal support for the habit and thus efforts should continue to limit exposure to such content.
- Advertising and promotion
- public policy
Statistics from Altmetric.com
Tobacco use is responsible for over 430 000 deaths a year in the US and remains a major source of preventable morbidity and mortality across the world.1 It is estimated that in the 21st century, tobacco use could result in as many as 1 billion deaths worldwide.2 Despite efforts to reduce adolescent tobacco use in the US, a recent study shows that nearly half (46%) of 12th graders have tried cigarette smoking and 22% have smoked at least one cigarette in the past 30 days.3 An important influence on youth smoking is exposure to smoking and tobacco-related content in popular feature length films.4 5 Adolescents are heavy movie watchers,6 who imitate the smoking of their favourite movie actors.7–9 Longitudinal studies have also found that early adolescents, ages 10–14,10 and children, ages 9–12,11 exposed to movie smoking were much more likely to start the behaviour. Furthermore, a review by Charlesworth et al12 found strong evidence that movie smoking promotes youth smoking initiation and a meta-analysis by Wellman et al13 that controlled for a wide range of confounding variables found that exposure to pro-tobacco media and marketing, including tobacco portrayal in films, more than doubled the odds of youth becoming tobacco users. Additionally, a 2008 US National Cancer Institute report concluded that the link between movie tobacco exposure and youth smoking initiation is causal.14
Given the important role of films in youth culture, reliable and valid assessments of trends in movie tobacco use should inform smoking prevention efforts. Indeed, tobacco portrayal in US made films is particularly critical because these movies are popular in the US and the world. In 2007, North American box office sales were $9.7 billion and foreign ticket sales from major studios were estimated at $9.4 billion.15 Thus, if tobacco content in popular films were on the increase, this would be a serious source of concern for tobacco control efforts.
Research on time trends in tobacco portrayal in films has produced seemingly contradictory results (see table 1). Our review of this literature includes publications indexed in a variety of search engines using the search terms ‘smoking’, ‘tobacco’, ‘movies’ and ‘films’. Studies were included if they reported on tobacco portrayal trends in (a) top-selling US films that (b) covered all film ratings and types (not subgroups) and that (c) spanned more than 5 years of time since 1950. As seen in table 1, three studies found increasing trends in tobacco portrayal, six reported no change and four studies had decreasing trends for various time periods from 1950 to 2008.
The inconsistent tobacco trends in table 1 may be attributable to differences in time periods covered, the density and sales rank of films sampled per year, and either weak or no intercoder reliability reported. The 3 studies that reported an increase in tobacco content tended to have small samples (5 or fewer films per year) taken from a relatively narrow (top 20) sales range with no evidence of reliability.16–18 The 6 studies that reported no overall trend either covered a small time period (11 or fewer years)19–22 or had small sample sizes (between 6223 and 110 films19 or weak20 23 or no reported reliability19 21 22 24a. The four studies that reported declines had large dense samples with strong reliability. However, they only covered the period from 1990 to 2007.
Our review indicates that no study with a wide time frame, sufficiently dense sample and high reliability has been published to justify conclusions about long-term tobacco portrayal trends in films. The present study was designed to fill this gap by sampling a broad range of films (top 30 grossing films per year) that were sampled densely (15 films per year) and that covered a longer time period (1950–2006) than previous studies.
In designing this content analysis, we relied on theories of media influence, such as Bandura's social cognitive theory of mass communication,25 which suggests that media portrayal of tobacco use can influence viewers by any of several routes. The most influential mechanism is by providing models of the behaviour. Such models show viewers how to use the product and to the extent that the behaviour is explicitly shown (without negative effects), encourage imitation. In addition, portrayal of any cues linked to smoking (brand logos, smoke filled rooms, ashtrays with lit cigarettes) could also serve as prompts to the audience that can reinstate motives to smoke or remind users of their habit. Finally, media representations can cultivate a picture of the world that informs users, either correctly or incorrectly, about the acceptance and prevalence of tobacco use in society.26 Hence, we assessed the explicit modelling of the behaviour as well as all tobacco-related cues in films. The most inclusive measure of portrayal was the overall presence of any tobacco-related content, including characters smoking or the presence of cigars, ashtrays or tobacco billboards. A second measure focused on the actual use of tobacco by main characters, such as purchasing, handling or smoking a cigarette, cigar or pipe.
To determine how closely media portrayal reflected usage and acceptance of tobacco in the culture, our analysis also examined long-term trends in adult per capita cigarette consumption and major landmarks in tobacco control efforts since 1950. Given the causal role of tobacco portrayal in youth initiation,14 we would expect historical trends in tobacco portrayal to parallel per capita consumption. However, tobacco control efforts should influence consumption and could potentially affect movie portrayal trends as well. Hence, it was not feasible to test for a causal relation between films and population consumption. However, these comparisons provided an opportunity to determine how well trends in film portrayal of tobacco reflect actual cigarette consumption trends. To the degree they run in parallel, films could serve as a barometer of societal trends in cigarette consumption and thereby provide viewers with a picture of cultural acceptance of smoking.
This study is part of a larger project to identify trends in media portrayal of health risks (see http://youthmediarisk.org). This study identified the 30 highest grossing films based on US box office sales for each year from 1950 to 2006 as compiled by Variety magazine. These represent about 50% of the total box office sales as estimated for recent years (2001–2006) for which reliable data were available. Starting with either rank 1 or 2, films from every other sales rank (15 per year for a total of 855) were selected for coding with the starting rank being randomly selected using a computer generated random number process. In cases where films were not available for purchase (about 5% of films), the next ranking film was substituted.
A total of 24 undergraduate students participated in the coding process. Training involved mastering an extensive codebook with rules for assigning codes to film content (see http://www.youthmediarisk.org for greater detail about the coding instrument). A set of about 21 h of diverse film content was used for training purposes. Coders were required to exhibit a high level of reliability before completing training. We used Krippendorff's α formula (Kα)27 to assess reliability. This formula controls for chance agreement and can be used for multiple coders employing either nominal (yes vs no) or rating scale codes. A value of 0.70 was used as the minimum acceptable level for any content category.
Films were coded in 5-min segments. Thus, a 2-h movie would have 24 segments. Three codes were assigned to each 5-min segment.
Total tobacco content
Segments that contained tobacco use by any background, minor, or main characters including non-human and all animated characters, or the appearance of any tobacco-related content, including smoking adverts, logos, references, or paraphernalia were coded as 1 (versus 0; Kα≥0.78). About 86% of films were identified with such content. The mean percentage of segments per film with total tobacco content was 31.6.
Main character tobacco use
While tobacco use by all major and minor characters was captured by the total tobacco measure, we also examined the portrayal of tobacco use by a maximum of eight main characters. This measure captured the actors that were most central to the plot and likely to have the greatest impact on the audience, including non-human and all animated main characters. Characters were identified using Halliwell's film guide,28 as well as the movie packaging and the Internet Movie Database (http://www.imdb.com), which often contains pictures of the actors. About half of the films had seven or fewer main characters. If a segment had a main character using tobacco, the segment was coded as 1 (vs 0; Kα≥0.82). The mean percentage of segments per film with a main character using tobacco was 16.3. About 69% of films contained at least one main character using tobacco.
Because we assessed any main character use during a 5-min segment, it was possible that we could overlook multiple occasions of tobacco use within a segment. Hence, we also assessed the intensity of tobacco portrayal per segment using a five-point scale (Kα≥0.81) (0=no tobacco or related content; 1=had tobacco or related content; 2=tobacco product present but not consumed; 3=one character consuming tobacco or smoking; 4=two or three characters consuming or smoking tobacco; 5=more than three characters consuming or smoking tobacco). However, because this measure was highly related to total tobacco content (r=0.83, p<0.001) and exhibited a very similar time trend, it is not discussed further.
Annual US per capita cigarette consumption was used to provide an historical picture of tobacco use. Data were taken from tobacco consumption reports by the US Centers for Disease Control and Prevention and the US Census Bureau.29–31
The coding unit was the 5-min movie segment. However, to control for film length, we calculated the percentage of each film's segments that contained total tobacco content or main character tobacco use. To identify trends in annual tobacco portrayal, we calculated the arithmetic and geometric mean of each measure for the 15 films per year. These scores were then plotted by year to assess trends. Although the 15 film scores per year tended to be skewed, there was little difference in annual trends between using the arithmetic or geometric mean. Hence, we present results for arithmetic means (N=57).
Because the trends in portrayal were adequately described by linear and quadratic terms, we used ordinary least squares regression to identify trends over the 57-year period. We also tested for the presence of autocorrelation in each series. No autocorrelation was found in either film measure of tobacco content. However, autocorrelation was observed in US cigarette consumption and was controlled in the analyses of this outcome.
There was a decline in total tobacco content (r2=0.820, p<0.001) described by a downward linear (b=−0.348, 95% CI=0.045, 0.652, p=0.025) and quadratic trend (b=−0.016, 95% CI=−0.021, −0.011, p<0.001). The model's best fitting trend started at 38.9% in 1950, reached a maximum of 40.8% in 1961 and dropped to 7.9% in 2006 (figure 1A). There was little change in the trend from 1950 to 1970; but beginning in 1971, it dropped at an increasing rate. The percentage of films with main characters using tobacco also declined (r2=0.711, p<0.001) described by a significant downward quadratic trend (b=−0.005, 95% CI=−0.009 to 0.000, p=0.027). The main character model's best fitting trend began at a maximum of 23.3% in 1950 and dropped to a minimum of 4.5% in 2006 (figure 1B).
US adult cigarette consumption also decreased (r2=0.991, p<0.001) with a model described by a significant downward linear (b=−15.089, 95% CI=−7.057, −23.121, p<0.001) and quadratic trend (b=−0.342, 95% CI=−0.572, −0.112, p=0.004) (figure 1C). This trend started at 3677 cigarettes per capita in 1950, reaching a maximum of 4096 in 1966 and then declining to 1338 in 2006. To compare trends in film portrayal and cigarette consumption, we plotted standardised predicted values for all three curves (figure 2). It is evident that total tobacco content more closely paralleled US consumption. However, tobacco film content peaked somewhat earlier than consumption (1961 vs 1966), and main character portrayal simply declined over the time period. Also shown in the figure are major milestones in tobacco control efforts. It is clear that since at least 1964, tobacco control efforts have increased in tandem with declining cigarette consumption and tobacco movie portrayal.
The US National Cancer Institute concluded that the link between exposure to the use of tobacco in films and youth smoking initiation was causal14; therefore identifying trends in tobacco content in top-grossing films is important for tobacco control efforts. This study was designed to clarify the inconsistent results in the literature by examining films over a long time period (1950–2006) with a dense sample covering a wide range of top-grossing films. It also maintained a high standard of reliability and examined tobacco content with portrayal measures that should be linked to smoking initiation and subsequent reinforcement of the habit. Consistent with the studies identified in table 1 that also employed rigorous methods, we found a strong decline in total tobacco portrayal that began in 1961 and a decline that was already underway in 1950 for tobacco use by main characters.
It is important to note that the inconsistent results in the literature have sometimes led reviewers to differing conclusions regarding trends in tobacco content in films. For example, Charlesworth et al12 found a downward trend from the 1950s that then increased in the 1990s. But for the post 1990 period, this review relied on Glantz et al's18 study which had small samples rather than relying on the Mekemson et al32 study, which sampled more films and found a modest downward trend for smoking content. In contrast, a 2009 World Health Organization report33 concluded that tobacco portrayal in films for the period 1999–2006 exhibited no change. However, this report relied heavily on the methodologically weak study (see Polansky and Glantz22 for more details; see also table 1) in drawing this conclusion and it is a very short period in which to find a trend. Future reviews of this literature should devote greater attention to the methodological strengths and weaknesses of the research.
The trend in total tobacco content appeared to closely match US adult cigarette consumption over the 57-year study period. Consumption began to decline in the 1960s as tobacco control efforts were increasing. These included but are not limited to the 1964 US Surgeon General's report documenting the dangers of smoking and the elimination of television adverts for cigarettes (1971) and others (figure 2).34
The simultaneous trends in tobacco portrayal in films and adult smoking can be explained in at least three ways. The drop in tobacco portrayal could have contributed to the drop in adult cigarette consumption. This would be consistent with studies showing that youth exposure to tobacco content in films predicts subsequent initiation of smoking. This causal path would require several years to show up in adult consumption trends, and indeed we found that total tobacco peaked 5 years before consumption began to decline. In addition, portrayal of main character smoking began to decline in 1950. Additionally, the accelerating decline in cigarette consumption could have influenced scriptwriters and directors to include fewer tobacco-related cues and less smoking behaviour in their films. This would be most consistent with film trends after 1966 when consumption began to consistently decline. Finally, the increasing tobacco control efforts beginning in the 1960s may have influenced both trends. For example, increasing tobacco control efforts may have made the social climate less accepting of tobacco use thereby reducing cigarette consumption and tobacco movie portrayal. The 1950 report in JAMA may also have helped begin the process of raising concerns about tobacco, an effect that began to register in main character portrayal.
Our historical analysis cannot distinguish these alternatives. However, given the remarkable similarity in the patterns of total tobacco and consumption trends, it is safe to conclude that trends in tobacco portrayal in films have served as a barometer of societal acceptance (or lack thereof) of tobacco use since 1950. However it is also important to note that while movie representation may be a barometer of smoking trends in society, it has not necessarily accurately reflected specific demographic smoking trends. For example, Hispanic movie characters have been found to smoke more than surveys suggest they do in real life.35 Despite these forms of inaccuracy, it is quite likely that successive cohorts of young people have been influenced by the declining popularity of smoking whether it is in films or real life. Additionally, adult smokers have been exposed to fewer cues that might elicit craving or elicit the desire to smoke. Thus, whether or not tobacco portrayal in films has directly led to reductions in adult smoking, it has likely served as an important indicator of the normative climate surrounding smoking, and this may have influenced the decline in tobacco use since the 1960s.
There are some limitations in this study. This study's use of 5-min segments for analysing tobacco content may have been less sensitive than a continuous measure of tobacco instances. Nevertheless, the trend we observed in total tobacco portrayal would have been very similar if we had also reported an intensity measure capturing all instances of smoking in a segment. Furthermore, the trends we have observed since 1990 are consistent with those reported in studies that used continuous measures,36 1991,32 and 199637 38 (table 1). Hence, there does not appear to be a lack of sensitivity for detecting important trends in onscreen portrayal. This study's observed downward trends in tobacco portrayal represent top 30 box office sales and may not generalise to less popular films. However, for the most recent years of the study the top 30 films account for a sizeable 50% of total domestic box office gross sales.39 40 Future studies should examine cult films, internet-based movie content and less popular films that may nevertheless appeal to young potential smokers.
In May 2007, the Motion Picture Association of America (MPAA) announced that ‘all smoking will be considered and depictions that glamorise smoking or movies that feature pervasive smoking outside of an historic or other mitigating context may receive a higher rating’ (page1).41 However, a 2008 report concluded that this effort had only produced minimal change in film ratings in its first year.42 A subsequent 2009 report said the MPAA did not raise the rating of any theatre-released films because of tobacco content.43 Thus, more seriously weighing movies with tobacco content or giving them all a restricted (R) rating is worth considering. A complementary strategy could involve showing anti-smoking adverts before films that contain smoking.44 45
Although this study finds a decline in tobacco portrayal in top-grossing films since 1950, tobacco portrayal in films remains problematic and tobacco control efforts should continue working to limit harm from tobacco-related content portrayed in films.
What this paper adds
Because exposure to the portrayal of tobacco use in movies has been causally linked to smoking initiation in young people, identifying tobacco content trends in movies is important for future tobacco control efforts.
Unlike other studies that covered fewer years and had less certain intercoder reliability, this study analysed and found downward trends in tobacco use by main characters in films since 1950 and by total tobacco content in films since 1961.
The decline in movie tobacco portrayal coincided with declines in tobacco consumption and an increase in major tobacco control efforts. However, despite this study's evidence that movie tobacco portrayal has declined since 1950, tobacco control policy should continue to promote strategies that minimise harm from portrayal of tobacco use in movies.
The authors thank the staff and coders of the Adolescent Risk Communication Institute's CHAMP project (see http://www.YouthMediaRisk.org), Dr James Sargent for his comments, the Annenberg Public Policy Center (APPC) of the University of Pennsylvania and the Robert Wood Johnson Foundation for their financial support of the CHAMP.
Funding Annenberg Public Policy Center, Robert Wood Johnson Foundation.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.