Cigarettes and celluloid: a dubious link

Patrick Basham and John Luik argue that the calls to ban smoking in films are based on dubious research.

One of the ‘nice’ things about the advocacy of the anti-tobacco lobby is how consistently silly, not to say nonsensical, its claims are. Unencumbered with the responsibilities of reputable research and rigorous analysis of whether their purported solutions really work, the anti-tobacco zealots are able to continue, year after year, promoting their claims as the novel product of serious thinking on the problem of preventing smoking.

Take, for instance, the increasing media attention – generated entirely by the anti-tobacco movement – to the claim that smoking in films is a central ’cause’ of youth smoking. The latest example in the UK is a report from the British Medical Association published earlier this month, Forever Cool: The Influence of Smoking Imagery on Young People, which argued that smoking on the small and big screens encouraged youngsters to take up the habit (see The BMA: censorious, busybody killjoys, by Tim Black).

There have been an unhealthy stream of such reports over the years. Anti-smoking activist Stanton Glantz has claimed that seeing onscreen smoking is the main reason why teens begin to smoke. Indeed, Glantz even provided precise figures on the effect in the Lancet in 2003, arguing that ‘smoking in movies is responsible for addicting 1,080 US adolescents to tobacco every day…’.

And Glantz is not alone. A report published in 2007 by the US Institute of Medicine, Ending the Tobacco Problem: A Blueprint For The Nation, wades into the smoking in films controversy by claiming not only that ‘exposure to smoking in movies increases the risk for smoking initiation’, but that this increased risk ‘can be reduced by antismoking advertisements’. The report recommends that the Motion Picture Association of America (MPAA) should consider tobacco use in films as a factor in assigning a film an R-rating (which bars those under 17 years old unless accompanied by an adult). Indeed, the Association does appear to be considering including tobacco usage in assessing film ratings.

Additionally, a study published in 2006 – actually merely a meta-analysis of previous studies – by University of Massachusetts researcher Robert Wellman has been widely reported as providing evidence of a ‘causal link’ between smoking in films and becoming a smoker.

India has already heeded these calls to eliminate on-screen smoking. Australia has taken the different and controversial approach of screening 30-second anti-smoking ads shot in movie-trailer style before films which contain scenes with smoking. The ads show actors smoking and warn viewers not to be sucked in by the ‘tobacco giants’ who ‘crave fresh blood’.

Unfortunately the ads have not had quite the effect that was intended. A study in Tobacco Control by Christine Edwards and others examining the Australian approach found that the advertisements did not produce reduced intentions to smoke. For instance, 25 per cent of smokers who had not seen the ads before films with smoking said that they would still be smoking in a year’s time compared to 39 per cent of smokers who had seen the anti-smoking ads. The ads had no statistically significant effect on the smoking intentions of nonsmokers. All of which suggests that the pre-smoking film ads are like so many anti-tobacco initiatives: counterproductive.

There are three reasons to doubt that films that show smoking cause youths to smoke. To begin with, the research – at least 10 major articles during the last five years – which purports to show this causal connection is deeply flawed. For example, the analysis by Wellman et al, published in December 2006, argues that films which depict smoking create positive views about smoking and these positive views lead to an increased intention to start smoking. To support this claim, Wellman analyses 51 studies with 89 exposure measures. But of these 89 measures of exposure, only 10 were about exposure to smoking in films – a very slender base on which to claim that such exposure causes adolescent smoking.

Further, though Wellman cites crucial studies about the huge number of things that influence adolescents to begin smoking – some studies suggest over 200 different risk factors – he notes that the studies he uses in his meta-analysis fail to take account of these other risk factors for smoking. Yet without taking account of these it is impossible to determine what influence, if any, smoking in films might have. Most crucially, he fails to note several studies which have consistently found that young movie viewers do not rate smoking characters as particularly attractive. This undermines his central thesis that depictions of smoking create a positive view about smoking. It seems rather a stretch to conclude that adolescents are driven to smoke by smoking characters whom they do no like.

Nor is the questionable quality of the Wellman research unique. An often-cited study, by Madeline Dalton in the Lancet in 2003, claims to establish a causal connection between smoking in films and youth smoking. Yet the researchers did not even ask their subjects whether they found a film’s portrayal of smoking to be positive – which is surely the relevant research question given the assumption that such a positive attitude is the cause of beginning to smoke. Instead, Dalton assumes that because her subjects started smoking after seeing someone smoking in a film that they must have started smoking because they saw someone smoking in a film.

Dalton then typifies the second reason why the claim that seeing smoking in films leads youth to smoke cannot be trusted: it is based on epidemiological studies that cannot, by their nature, ever establish such causal claims. Such studies are always merely associational. Dalton and Wellman claim to skirt this limitation through prospective studies in which nonsmokers are studied. Both tell us that because the kids they studied only started smoking or at least experimenting with smoking AFTER they saw movies with people smoking, then we can be certain that they started smoking because they saw movies with people smoking.

This, however, is both confused and dishonest. Sequence is not consequence as anyone who has ever studied logic will remember. Post hoc ergo propter hoc – ‘after this therefore because of this’ – is still a fallacy, even in anti-tobacco land. Just because something follows something else – in this case, starting to smoke follows seeing movies with smoking scenes – it is by no means necessarily true that the one is the cause of the other.

The serious research literature on the causes of youth smoking has discovered dozens of risk factors. Wellman even acknowledges that there are a variety of ‘psychosocial factors’ which increase the odds of starting to smoke. Yet Dalton controls for only a handful of these even while claiming that viewing on average 17 films in which there is smoking leads young people to begin smoking.

There is an equally plausible interpretation of these and similar results. Since smoking scenes occur most often in R-rated films, it could be that young people who view such films are more likely to become smokers, regardless of the content of the films. There is an association between R-rated films and smoking, but it is not a causal one. The things which the smokers have in common is their attraction to R-rated films, not to the smoking scenes in R-rated films. In other words, potential youth smokers share a number of characteristics such as liking R-rated films, but this does not mean that R- rated films with smoking scenes lead to youth smoking.

Alternatively, risk factors for smoking other than smoking scenes might be found in R-rated films. This possibility is in fact acknowledged by Sargent et al in their study on smoking in films and youth uptake, but dismissed on the odd grounds that their explanation is more ‘theoretically reasonable’. So indeed was the belief that the sun revolved around the earth. Unfortunately, the science suggested otherwise, as it does here.

But there is a third reason why such claims about films with smoking leading to smoking initiation cannot be trusted: there is a very substantial mass of research which not only points in different directions but contradicts these claims. For one thing, there are dozens if not hundreds of studies which find that the strongest predictors of youth smoking initiation are personality variables such as rebelliousness, socioeconomic factors such as family income, school-related factors such as academic success and remaining in school and such coping skills as resilience. These make nonsense of the Glantz claim that seeing onscreen smoking is the ‘main’ reason why teens begin to smoke.

For another thing, the research about the supposed effect of smoking in films on youth smoking itself contradicts the claim that seeing smoking leads to smoking. The study from Australia about the ineffectiveness of showing anti-smoking ads prior to films with smoking notes that the young people’s views about smoking were linked to their current smoking status, age and gender but ‘not to the movie viewed’. In other words, the smoking scenes in the films these young people viewed were NOT responsible for their beliefs about smoking – or, as the study shows, their intention to smoke in the future. So the very sort of quasi-experimental study that is so praised by anti-tobacco activists actually refutes their core claim that smoking scenes lead young people to start smoking.

Finally, there is the odd claim made by Wellman that even if the evidence for a causal connection between seeing smoking in films and starting to smoke is weak, we know it is true because there is a causal relationship between violent media such as video games and ‘violent attitudes and behaviors in children and adolescents’ which ‘makes it plausible that tobacco marketing and media also affect behavior’. Aside from the fact that only one study is listed as support for this sweeping conclusion about video games and violence, and that virtually all of the research is correlational and hence cannot establish causal connections, there is little evidence supporting such links between video games and violent behaviour according to the Federal Trade Commission’s review of the issue.

Youth smoking has suffered from over 25 years of bad research and simplistic policy prescriptions from the anti-tobacco movement. Despite the fact that the evidence is not supportive – there is no causal connection between tobacco advertising bans and reduced youth smoking, for instance – the anti-tobacco activists have always claimed that advertising is the major cause of young people starting to smoke. Now with tobacco advertising increasingly banned and young people still smoking, the claim is that it is seeing smoking in films that is the major cause of young people beginning to smoke. But the evidence for this is just as contrived and flimsy as it was for the claims about tobacco advertising.

Removing smoking from films, or young people from cinemas that show films with smoking, will not do anything to solve the problem of youth smoking. But it will sidetrack attention from the consideration of policies that might.

Patrick Basham and John Luik are authors of a forthcoming book, Up to Snuff, to be published by the Social Affairs Unit, examining the negative consequences of the ban on smokeless tobacco in the European Union.