Data were derived from a survey of respondents drawn from a nationally representative Web-based sample of 748 Korean high school students (376 males and 372 females) who participated in the Hankook Research Master Sample Panel. Respondents were recruited using a dual sampling frame. A combination of random digit dial and address-based sampling allowed for the selection of a sample of individuals without using telephone land lines. Respondents received a nominal cash incentive (US $2.50) to participate when they completed both the baseline and post-movie surveys. The final response rate was 81.0%. Missing values of survey questions for key analytical variables were excluded using the pairwise method.

The students (aged 16-18 years) were randomly assigned to watch three professionally edited short (40-120 seconds) video clips of representative popular Korean films with smoking scenes (exposure group, n=374) or without smoking scenes (non-exposure group, n=374). Before the participants started the Web-based survey, the statistician randomly assigned each of them to two different study groups (exposure/non-exposure), according to a table of computational random numbers. The random assignment was blocked by gender to distribute males and females equally in each group. The video clips were professionally edited versions of representative hit films from among 51 films that played in Korean cinemas from 2000-2013 [19]. We defined a hit film as a movie that had been viewed by at least 5 million people [19]. The participants watched the three video clips in a row by themselves while completing the Web-based survey. The participants were unable to skip watching certain videos. In the exposure group, all scenes presented contained images of smoking. Participants’ desire to smoke was gauged immediately after viewing the clips. We examined whether smoking scenes in motion pictures could increase adolescents’ smoking desire. We also examined whether adolescents’ smoking desire could be promoted or restrained by predisposing conditions, such as sensation-seeking or health information-seeking behavior. After adjusting covariates using propensity score matching (PSM), we conducted a paired-sample t test to compare the smoking desire of respondents before and after viewing smoking scenes.

Survey questions were based on previous reports on the effects of social context, including mass media use and social capital, on health [17,19,20]. The questionnaire included topics adapted from the Health Information National Trends Survey of validated measures developed through a series of health communication studies in Korea [12,21]. Sensation-seeking behavior was defined as an individual’s innate propensity to seek out novel, strong, and intricate experiences and feelings and to willingly take physical and socioeconomic risks to have such experiences and feelings [22,23]. Sensation-seeking is strongly associated with a variety of illegal and risky behaviors, such as smoking, among adolescents [24]. The Cronbach alpha values of all individual subscales ranged from .60 to .87 in this study.

Results of a cross-tabulation analysis conducted to verify whether the experimental group and the control group were well matched are shown in Tables 1 and 2. For male respondents, the multivariate imbalance measure was 0.162 before matching and decreased to 0.111 after matching, indicating good matching. Male respondents with a “very negative” opinion of smoking comprised 60.5% of the control group and 56.8% of the experimental group. Concerning the effect of smoking on health, 67.3% of respondents in the control group and 59.3% of respondents in the experimental group believed smoking was fatal. Similar percentages of both groups responded that one or both of their parents smoked or that both were non-smokers.

For female respondents, the multivariate imbalance measure decreased from 0.246 before matching to 0.209 after matching. The proportion of females with a very negative attitude to smoking was 63.4% in both the control group and the experimental group. Similar percentages of females believed that smoking is fatal (66.0% and 68.6% in the control and experimental group, respectively). Concerning parental smoking, 54.9% of the control group responded that both parents were non-smokers with 56.2% of the experimental group responding that one or both smoked.

Table 3 presents the data concerning cigarette cravings between the experimental and control groups based on independent sample t tests. Cigarette craving was significantly higher in the experimental group than that in the control group among male adolescents (t_307.96=2.066, P<.05; Table 3). In the experimental group, cigarette cravings of adolescents after viewing smoking scenes were significantly higher than they were before watching the video clips (t_161.00=2.867, P<.01; Table 4). Figure 1 depicts the results graphically.

Logistic regression was performed through PSM to determine factors associated with changes in cigarette cravings among the male high school students following viewing of smoking scenes in motion pictures. Results are summarized in Tables 5 and 6. Table 5 provides descriptive statistics of the sample. According to Model III, the video clips did not significantly influence cigarette craving (Table 6). However, more impulsive adolescents had significantly higher cigarette cravings than less impulsive adolescents: adjusted Odds Ratio (aOR) 3.40 (95% CI 1.40-8.23). On the contrary, cigarette cravings did not significantly increase in adolescents who habitually engaged in health information-seeking. In particular, those who very actively sought health information had considerably lower cigarette cravings than the group who did not engage in such information-seeking: aOR 0.08 (95% CI 0.01-0.88).

This study revealed a direct link between viewing smoking scenes and immediate subsequent smoking desire. In particular, male adolescents who watched smoking scenes were more likely to have the desire to smoke than those who watched scenes that did not contain smoking. This is consistent with findings in the literature that people 18 to 25 years of age are influenced by smoking scenes in movies and are more apt to initiate smoking [31,32]. Presently, cigarette cravings in male high-school students were markedly increased in those displaying impulsive behavior, which is a known risk factor for smoking. The urge to smoke following viewing of the smoking scenes was considerably less in those who habitually engaged in health information-seeking behavior. The former group perceived tobacco as a substance of curiosity. In contrast, the latter group had negative attitudes and beliefs regarding tobacco, which were reinforced by information-seeking.

One-third of smokers start smoking when they are adolescents. Failure to quit smoking during adolescence considerably increases the chance of becoming a long-term smoker [33]. Smoking scenes frequently appear in movies rated PG-15. This makes it very likely that Korean adolescents will view the scenes. In impressionable individuals, the exposure could be an important factor in their decision to smoke. One factor could be a reinforcement of their burgeoning positive attitude towards cigarettes. Appropriate regulations and management are necessary to curb this influence. Banning smoking scenes in films is seemingly an obvious option. However, filmmakers counter that dramatic necessity and freedom of expression can allow for depiction of smoking, especially when it is deemed essential in the storyline [34].

Some countries, such as the United Kingdom, have banned the promotion and advertisement of tobacco. However, they still allow movies with positive messages towards tobacco to be viewed. Many R-rated movies produced in the United States from 2001 to 2006 had smoking scenes and 90% of viewers of such movies reported that they had a greater urge to smoke after watching these movies [35,36]. However, more people agree that regulations against smoking scenes in movies are needed. Positive changes to regulate smoking scenes in movies are being made. In the United States, Hollywood film companies are practicing self-regulatory rules to ban scenes where protagonists smoke in G, PG, and PG-13 movies [37]. This decision was based on research suggesting that the more R-rated movies a minor watches, the more likely it is that this person will smoke in the future [32]. India, another country with a robust film industry (“Bollywood”) is also strengthening regulations on smoking. Tobacco brands were frequently shown in Bollywood films made in Hindi until the early 2000s [38]. However, as the government espoused the reduction of smoking scenes in movies, vetting of films to decide whether the smoking scene is necessary has been instituted [39]. To increase the effectiveness of the ban on smoking scenes, the Indian government is allowing smoking scenes but only if the scene is out-of-focus and accompanied by a 30-s message in the film warning of the danger of smoking [40].

Establishing a standard that restricts the frequency of smoking scenes in films and assigning a smoking-related screening grade to films are warranted. Our findings suggest that adolescents who attempt to quit smoking should lessen or refrain from viewing motion pictures that contain smoking scenes.

Some study limitations should be noted. First, the number of friends who smoke is an important covariate, but this study was not able to include any type of information on social networks among adolescents in the questionnaire due to ethical problems. However, regarding adolescent smoking, the significance of viewing smoking scenes in motion pictures may be amplified by social networks rather than attenuated [8]. Second, data on smoking desire were collected via a self-reported questionnaire. So, the results may have been subject to recall bias. In addition, the Hawthorne effect may have occurred because participants knew the purpose of the study. Future research needs to develop a longitudinal study and an investigation and management system to conclusively establish the reliability and validity of the data.

This study reveals a direct link between viewing smoking scenes and immediate subsequent smoking desire. In particular, male adolescents who watched smoking scenes were more likely to have the desire to smoke than those who watched scenes without smoking and were more apt to commence smoking. Cigarette craving increased considerably in male high-school students with an impulsive, risk taking personality. However, the desire to smoke cigarettes was considerably decreased in those who habitually engaged in health information-seeking behavior. Thus, a comprehensive effort involving dissemination of information is needed to promote global health. Effort is especially needed to prevent adolescents from being exposed to smoking scenes in movies. First, we need to identify the more vulnerable adolescents who are being exposed to smoking scenes in films. At the same time, we need to consider a targeted approach by finding predisposing conditions, such as sensation-seeking or impulsivity. Health campaigns that foster health information-seeking among young people are a good idea. Second, rather than implementing downstream and selective protection plans targeting adolescent viewers and their parents, an upstream approach is needed. For example, if the movie industry is genuinely interested in this project, civil servants, movie creators, and academics can talk together to understand the necessity of rules against smoking scenes and find ways to come up with appropriate solutions.