Background Cigarette pictorial warning labels (PWLs) could produce stronger quit intentions than text-only warning labels (TWLs) due to greater emotional arousal. Yet, it remains unclear whether PWLs that elicit different levels of emotions produce different outcomes. To better understand the role of negative emotions in the effects of PWLs, this study developed two sets of PWLs arousing different emotional levels (high vs low) but equally high on informativeness and compared them to each other and to the current TWLs.
Methods Adult US smokers (n=1503) were randomised to view nine high-emotion-arousing or low-emotion-arousing PWLs or TWLs. After each label, participants reported the negative emotions they felt while looking at the label. After seeing all the labels, participants reported their intentions to quit smoking. Mediation analyses tested whether message condition influenced quit intentions indirectly through negative emotions.
Results Compared with TWLs, PWLs produced higher levels of negative emotions (b=0.27, SE=0.04, p<0.001). Compared with low-emotion arousing PWLs, high-emotion-arousing PWLs produced higher levels of negative emotions (b=0.24, SE=0.07, p<0.001). Higher negative emotions predicted stronger quit intentions (b=0.20, SE=0.03, p<0.001). Negative emotions mediated the effects of PWLs versus TWLs and high-emotion-arousing versus low- emotion-arousing PWLs on quit intentions.
Conclusions The results provide additional evidence for negative emotions as the mechanism through which PWLs motivate smokers to consider quitting. The findings call on the Food and Drug Administration to design and implement high-emotion-arousing cigarette warning labels.
- cigarette warning labels
- pictorial warning labels
- quit intentions
Statistics from Altmetric.com
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.
Tobacco use is the leading cause of preventable death and disease worldwide, accounting for more than 7 million deaths globally each year.1 To educate people about the health risks of smoking, most countries have implemented warning labels on tobacco products. Over the years, these labels have evolved from brief text statements to graphic images with elaborate textual messages.2 As of 2016, over 100 countries have required pictorial warning labels (PWLs) on cigarettes.3 In 2009, the US Congress mandated the Food and Drug Administration (FDA) to implement PWLs for cigarette packages. The first set of nine PWLs was developed based on past research4 and the FDA-commissioned study that tested 36 potential versions of the labels.5 In explaining the selection of the final PWLs, the FDA highlighted the role of negative emotional reactions as indicators of message effectiveness.4 Tobacco companies in their judicial challenges argued that the selected PWLs violated their First Amendment (free speech) rights because they ‘were chosen not to convey information, but to evoke negative emotions and thereby discourage smoking’.6 The court sided with tobacco companies, finding that the PWLs were ‘unabashed attempts to evoke emotion’.6
In its decision, the court distinguished between ‘factual’ text-only warning labels (TWLs) and ‘emotional’ PWLs, claiming that while TWLs inform consumers, PWLs do not provide ‘purely factual and uncontroversial’6 information and instead ‘browbeat consumers into quitting’.6 Since the court’s 2012 decision, multiple public health and legal scholars have argued and supplied empirical evidence that, although PWLs evoke more negative emotions than TWLs,7 8 both TWLs and PWLs evoke negative emotions, both types of labels are viewed as factual and informative7 and that ‘emotional’ and ‘factual’ are not mutually exclusive concepts.9–11
Emotions play an important role in human decision-making.12 In the context of antismoking education, greater negative emotions evoked by antismoking messages can motivate stronger quit intentions.13 14 This could be explained from three theoretical perspectives.10 15 First, negative emotions can serve as an information heuristic. Having felt negative emotions while looking at cigarette warning labels, people might be more likely to view smoking as a risky behaviour16 17 and hence report stronger quit intentions (‘affect as information’).18 Second, negative emotions could act as a motivation and induce greater behavioural tendency to avoid a risk, that is, produce stronger quit intentions in the context of antismoking communication (‘affect as motivation’).19 20 Third, negative emotions may stimulate people to scrutinise the warning labels, increasing the warnings’ perceived credibility, which ultimately leads to heightened quit intentions (‘affect as spotlight’).21
Most research supporting the effectiveness of emotional antitobacco messages compared pictures with text,22–24 and only a few studies compared effectiveness among different types of PWLs, for example, by looking at specific features of PWLs, such as depiction of body parts or symbolic representations.25 26 As a result, little information is available on whether it is the presence of the picture itself or the emotions aroused by the image that serve as the mechanism behind the effects of PWLs. Therefore, research is needed on the effect of emotions that people experience while exposed to the different PWLs, going beyond the superficial segregation of informational and emotional messages as the distinction between text and image.
Two recent studies examined the effects of PWLs selected for low and high emotional arousal.15 27 They found that high-emotion-arousing PWLs resulted in stronger quit intentions, and low-emotion arousing PWLs resulted in lower quit intentions compared with TWLs.15 27 While these studies varied the levels of emotions, the two sets of PWLs likely differed on other features, such as perceived informativeness of the labels. Past research found that emotionality and informativeness of warning labels are highly correlated.7 Higher informativeness of warning labels was associated with higher odds of expressing antismoking attitudes, beliefs and behavioural intentions.28 Therefore, informativeness of warning labels might confound the findings on the effects of emotions in response to PWLs.
Our study aimed to address this issue by controlling for warnings’ informativeness when examining the role of emotions in the effects of PWLs on quitting intentions. Because high-emotion-arousing and low-emotion-arousing PWLs are defined based on individuals’ perceptions, we followed Tao and Bucy’s call29 and employed a mediation model of experimental research to test effects of PWLs with different levels of emotions. Specifically, the mediation model tests the role of emotions both as a message attribute (an intrinsic message property) and as a psychological state (participants perceptions). Both need to be evaluated in a single model to avoid ‘conflating two different classes of variables and […] assuming uniform responses to messages.’29 O’Keefe also noted that when message properties are defined in terms of individuals’ perceptions, testing individuals’ perceptions (emotion arousal, in this study) as a mediating mechanism of the message effects produces conceptual consistency.30 We focus on negative emotions as a mediator because past studies showed that negative emotions are the mechanism through which pictorial warning labels influence quit intentions.31 32
We hypothesised that (1) compared with TWLs, PWLs would evoke higher levels of negative emotions, which in turn would predict greater intentions to quit smoking; (2) compared with low-emotion-arousing PWLs, high-emotion-arousing PWLs would evoke higher levels of negative emotions, which in turn would predict greater intentions to quit smoking.
This study is part of a larger experimental study on PWLs. Participants were 2700 US individuals aged 18 years old and older, recruited purposively based on age and smoking status by Toluna (www.toluna-group.com), a survey market research company that enlists participants for their online research panel using multiple recruitment strategies, such as web banners, website referrals, affiliate marketing and pay-per-click. The total study sample consisted of non-smokers (never smoked or smoked less than 100 cigarettes in their lifetime, n=900), current smokers (smoked at least 100 cigarettes in their entire life and were currently smoking cigarettes every day or some days, n=1503) and former smokers (had quit smoking within the past 2 years, n=297). Because the study focused on understanding label effects on intentions to quit smoking, only data from current smokers were used (n=1503).
The Family Smoking Prevention and Tobacco Control Act (FSPTCA) specified nine warning statements to appear on cigarette packs.33 Rather than using the first set of nine PWLs developed by the FDA, which were struck down in court, we opted to develop our own labels to allow us to more clearly manipulate high emotions versus low emotions while keeping the informativeness level consistent and to provide new labels as the basis for the development of the next round of PWLs by the FDA. In collaboration with a social marketing company OnBeyond (www.onbeyond.com), we developed four PWLs for each of the nine warning statements. The PWLs were shown on the front of cigarette packs, covering 50% of the front surface, as mandated by the FSPTCA.33 For the control condition, we used the TWLs currently shown on cigarette packages in the USA. TWLs were shown on the sides of cigarette packs just as how they are currently presented in the USA (for details, see online supplementary file).
The FSPTCA does not prohibit the inclusion of additional information on the PWLs, and we provided additional facts on the labels to further facilitate consumers’ understanding of the risks of smoking. For example, the warning ‘Cigarettes cause cancer’ was amended to include ‘80% of lung cancer patients die within 3 years’. These statements had been tested in a prior study7 and were found to differ on the levels of negative emotions evoked.
To select high-emotion-arousing and low-emotion-arousing PWLs for the experimental study, 300 US adults (52% smokers, 48% non-smokers) evaluated to what extent these labels were informative and factual and the levels of negative emotions they felt while looking at them (measures are described below). Based on these evaluations, for each of the nine warning statements, we selected two PWLs that differed in the levels of negative emotions (high and low negative emotions) but were similar in perceived informativeness. Pilot tests showed that the levels of informativeness did not differ significantly between high-emotion-arousing and low-emotion-arousing PWLs (Mhigh=6.8 vs Mlow=6.8 on the 1–9 scale, p=0.76), but the levels of negative emotions were significantly higher in the high-emotion-arousing condition than the low-emotion-arousing condition (Mhigh=4.8 vs Mlow=4.4 on the 1–9 scale, p<0.001).
In an online randomised controlled experiment, participants began by reporting demographic information and past tobacco use. Then, they were randomly assigned into one of three conditions: (1) high-emotion-arousing PWLs, (2) low-emotion-arousing PWLs and (3) TWLs (control condition). In each condition, participants viewed nine PWLs in a random order. In the control condition, since there are only four TWLs currently used in the USA, participants saw each label twice plus one of the four TWLs selected randomly to ensure nine label exposure in all conditions. Participants were not limited to how long they could view each label but once they moved on, they could not return to view the same label again. After viewing each label, participants reported their negative emotions and perceived label informativeness. After seeing all nine labels, participants reported their intentions to quit smoking.
For each label, participants reported how much they felt afraid, sad, angry, guilty, disgusted and worried on a 9-point scale (1=not at all, 9=extremely).5 34 Responses to each emotion and each warning label were averaged into a composite score (α=0.97, M=4.78, SD=2.52).
Participants reported their intentions to quit in the next month on a 0 (very definitely no)−10 (very definitely yes) ladder (M=6.07, SD=3.30).35
Covariates included sex, race, education, other tobacco use status, perceived risk of smoking and nicotine dependence (Heaviness of Smoking Index, 0–6 scale, M=2.23, SD=1.44).35 36 Specifically, to determine other tobacco use status, we asked participants about ever and current use of moist snuff, snus, electronic cigarettes, hookah, little cigars or cigarillos and heat-not-burn products. Participants who had ever used any of these products but did not use in the past 30 days were classified as former users of other tobacco products, and those who reported using in the past 30 days were classified as current users of other tobacco products. The rest were classified as never users. Perceived risk of smoking37 was measured by asking: ‘Imagine that you just began smoking cigarettes every day. What do you think your chances are of having each of the following happen to you if you continue to smoke cigarettes every day? (a) lung disease, (b) lung disease other than lung cancer (such as COPD and emphysema), (c) heart disease, (d) early/premature death.’ Answers were on a 0 (no chance)−6 (very good chance) scale plus a response category ‘I don’t know’. In the data analyses, ‘I don’t know’ answers (n=42, 2.8%) were treated as missing. The other items were averaged into an overall index (α=0.93, M=4.72, SD=1.46). We also asked and controlled for participants’ pretest intentions to quit smoking: ‘(1) never expect to quit, (2) may quit in the future, but not in the next 6 months, (3) will quit in the next 6 months, (4) will quit in the next month, (5) are currently trying to quit’. Due to the small number of people selecting category 4, response categories 3 and 4 were combined into one category.
We also controlled for individuals’ perceived label informativeness. This variable might vary substantially across individuals even when controlled for across message conditions and hence, be a confounder of the influence of emotions on quit intentions.30 Perceived label informativeness was assessed following the presentation of each label through five items: ‘the label was informative’, ‘the label gave me a better understanding of the consequences of smoking’, ‘the label was based on facts’, ‘the label presented something that happens in real life’ and ‘the label portrayed an actual risk of smoking’. Participants responded on a 9-point scale (1=not at all, 9=extremely; α=0.97, M=6.53, SD=2.10). Means and SD of key measures by message conditions can be found in the online supplementary file.
We used Tao and Bucy’s mediation model of experimental research to test label effects.29 This model includes warning label conditions as an independent variable, emotional response of participants to the labels as a mediator variable and quit intentions as an outcome variable. The mediation model allows us to overcome conceptual and research design limitations of the models that test either the effects of the messages (message attributes) or the effects of participants’ perceptions of them (psychological states). Message condition was recoded into two dummy variables. One dummy variable compared high-emotion-arousing to low-emotion-arousing PWLs and the other compared high-emotion-arousing and low-emotion-arousing PWLs to the TWLs.
We used Hayes’ PROCESS macro V.3.338 model 4 to test whether message condition influenced quit intentions indirectly through levels of negative emotions. Significance of the indirect effect was determined by observing the 95% bias-corrected CI of the effect obtained through bootstrapping with 5000 resamples and associated p values. Indirect effect was regarded as significant when the CI did not contain zero.38 Hayes’ PROCESS model 4 only allows testing the mediation effects of one independent variable at a time. Given we had two message dummy variables, we ran the same model twice. Each time, one dummy variable served as an independent variable and the other dummy variable was controlled for. We tested the two mediation effects separately. Analyses were performed in SPSS V.24 and included the covariates mentioned above. Significance level was set at p<0.05.
Participants were 47% female, 66% White and 43.2% college graduates; 56% currently used other tobacco products and 42% reported currently trying to quit smoking (table 1). Perceived label informativeness correlated highly with negative emotions (Pearson r<0.61, p<0.001).
Our first hypothesis predicted that compared with TWLs, PWLs would evoke higher levels of negative emotions, which in turn would predict greater intentions to quit smoking. This was supported (figure 1). Compared with TWLs, high-emotion-arousing and low-emotion-arousing PWLs produced higher negative emotions (b=0.27, SE=0.04, p<0.001) and in turn stronger quit intentions (b=0.20, SE=0.03, p<0.001). The indirect effect was significant, b=0.06, SE=0.01, 95% CI=0.03 to 0.08. Controlling for negative emotions, the direct effect of high-emotion-arousing and low-emotion-arousing PWLs (vs TWLs) on quit intentions was not significant (b=−0.06, SE=0.05, p=0.18).
Our second hypothesis predicted that compared with low-emotion-arousing PWLs, high-emotion-arousing labels would evoke higher levels of negative emotions, which in turn would predict greater intentions to quit smoking. This was also supported (figure 2). Compared with low-emotion-arousing PWLs, high-emotion-arousing labels produced higher levels of negative emotions (b=0.24, SE=0.07, p<0.001), which in turn produced stronger quit intentions (b=0.20, SE=0.03, p<0.001). The indirect effect was significant, b=0.05, SE=0.02, 95% CI=0.02 to 0.08. Controlling for negative emotions, the direct effect of high-emotion-arousing (vs low) emotion-arousing PWLs on quit intentions was not significant (b=−0.14, SE=0.08, p=0.07).
PWLs are effective in educating people about the health risks of cigarettes and in encouraging them to quit smoking.8 15 39 40 Emotions evoked by the warnings seem to be one of the central mechanisms of these effects.31 32 However, past research on the effects of emotional PWLs has been limited because, like the US courts, it frequently equated emotional messages with PWLs and rarely examined the differences in emotions within different types of PWLs.27 40 Our study evaluated the role of emotions in the effects of PWLs on quit intentions. The two sets of PWLs we developed for this study evoked different levels of emotions, but both were consistently perceived as informative and factual.
The current study showed that compared with TWLs, PWLs increased negative emotions, which in turn led to greater intentions to quit. We further demonstrated that PWLs that elicited higher levels of negative emotions (high-emotion-arousing labels) led to greater quit intentions compared with low-emotion-arousing PWLs via an indirect effect. This finding aligns with previous studies demonstrating that emotional reactions function as a behavioural motivation, directly increasing smokers’ quit intentions.19 20 41
This research has implications for tobacco regulations. After the original nine PWLs proposed by the FDA were struck down in court, the FDA set out to conduct further studies to develop a new set of PWLs.42 In 2016, public health and medical groups, along with individual paediatricians, sued the FDA to compel the agency to issue a final rule requiring cigarette PWLs, as stipulated by the Tobacco Control Act. In March 2019, the court gave the FDA a deadline of 15 April 2019 to finish all studies and of 15 March 2020 to issue the final rule on the new PWLs.42 When the FDA issues the final rule with the new PWLs, it is likely to encounter the same legal challenges from the tobacco companies. To address them, the FDA will need to demonstrate that the new PWLs are ‘purely factual and uncontroversial’, a concept that has not been clearly defined by the courts. Yet, this can be achieved by using unaltered or unmanipulated images depicting accurate information about actual consequences of tobacco use. In addition, demonstrating that the selected labels are viewed by actual consumers as factual will bring further empirical footing to the issue of what makes a label ‘uncontroversial’ and move beyond relying on judges’ oblique conjectures.9
In selecting the final PWLs, the FDA might be tempted to choose PWLs that are lowest on emotions to stave off the judges’ definitions of emotional pictures as being controversial. However, as this study and past research show,27 this might result in less effective PWLs, undermining the goal of reducing smoking rates. We argue that the FDA should develop factual labels and select those that are perceived as the most informative and factual. Furthermore, because emotions and informativeness are highly correlated,7 it is likely that those very images selected for their factual worth will also be highly emotional, since emotion is important in communicating factual information. In this study, where both high-emotion-arousing and low-emotion-arousing PWLs were highly informative and factual, using higher emotional PWLs led to greater quitting intentions among smokers by evoking higher levels of negative emotions. In selecting the final PWLs, the FDA should not shun high-emotion-arousing PWLs.
The results of this study should be interpreted with caution due to the following limitations. We did not measure actual quitting behaviours. Future work can replicate our studies by using behaviour as an outcome variable. Furthermore, our study used one-time brief message exposure, and the message effects were assessed immediately following message exposure. Future work could use a longitudinal design to better understand the effects of high-emotion-arousing versus low-emotion-arousing PWLs over time. Moreover, while the study sample had a diverse, heterogeneous demographic background, the sample was not probability based. We only examined adult smokers, many of whom were currently trying to quit, which limits the generalisability of the findings. Future studies should investigate how PWLs evoking different levels of emotions predict adolescent smokers’ quit intentions and influence non-smokers’ risk perceptions. For some PWLs, different information was presented in high-emotion-arousing and low-emotion-arousing conditions, which might partially account for our findings. While this study focused on negative emotions, future research should examine the role of positive emotions in response to pictorial warnings on tobacco products and explicitly compare the roles of positive and negative emotions in message effects.
In conclusion, the current study adds to existing research on the emotional mechanisms underlying the effect of PWLs on quit intentions. The results demonstrated that negative emotion evoked by PWLs is a unique factor explaining the effectiveness of PWLs. Moreover, controlling for perceived informativeness, high-emotion-arousing PWLs were more effective than low-emotion-arousing PWLs in encouraging smokers to quit smoking through negative emotional arousal. The findings call on the FDA to design and implement high-emotion-arousing PWLs.
What this paper adds
Negative emotions evoked by pictorial warning labels (PWLs) are highly correlated with perceived label informativeness, which may confound the understanding of the role of emotions in processing PWLs.
This study extended prior work on PWLs by differentiating between high-emotion-arousing and low-emotion-arousing PWLs while keeping the levels of perceived informativeness consistently high.
Controlling for perceived informativeness, high-emotion-arousing PWLs evoked more negative emotions and, in turn, higher quit intentions than low emotion-arousing PWLs.
Presented at This study was presented in American College of Surgeons – Michigan Chapter (2019).
Correction notice This article has been corrected since it was published Online First. An incorrect Acknowledgement section has been removed.
Contributors YL wrote the first draft. BY conducted the statistical analyses and wrote the method and results sections. DO wrote the method section. LP designed the study, developed the materials and wrote the implications. All authors contributed to the writing and revision and approved the final version of the manuscript.
Funding Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health and the Food and Drug Administration (FDA) Center for Tobacco Products (R00CA187460) and the National Institute of Drug Abuse of the National Institutes of Health and FDA Center for Tobacco Products (R01DA047397). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the FDA.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval This study was approved by the Georgia State University Institutional Review Board (H17198).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.
Author note This study was conducted while YL and BY were at Georgia State University.