Article Text
Abstract
Objective This study sought to demonstrate causal effects of exposure to Natural American Spirit (NAS) advertising content on misinformed beliefs of current and former smokers, and to empirically establish these beliefs as a mechanism driving intentions to use NAS.
Methods Our study employed a randomised experimental design with 1128 adult daily, intermittent and former smokers. We compared participants who were exposed to NAS advertisements or claims made in the advertisements with those in a no-message control group to test the effects of NAS advertising content on inaccurate beliefs about NAS and attitudes and intentions towards the product.
Results One-way analysis of variance revealed that exposure to NAS advertisements produced inaccurate beliefs about the composition of NAS cigarettes among current and former smokers (p<0.05). Planned contrasts indicated a compilation of arguments taken directly from NAS advertisements resulted in significantly greater beliefs that NAS cigarettes are healthier/safer than other cigarettes (for former smokers, t(472)=3.63, p<0.001; for current smokers, t(644)=2.86, p=0.004), demonstrating that suggestive claims used in the brand’s marketing have effects on beliefs not directly addressed in the advertisements. Regression and mediation analyses showed that health-related beliefs predict attitudes towards NAS for current and former smokers, and mediate intentions to use NAS.
Conclusions The findings of this study provide causal support for the need for further regulatory action to address the potentially harmful ramifications of claims used in NAS advertising.
- advertising and promotion
- tobacco industry
- public policy
- public opinion
- media
Statistics from Altmetric.com
Introduction
In 2015, the US Food and Drug Administration (FDA) issued a warning letter to Santa Fe Natural Tobacco Company (SFNTC) for the use of words such as ‘natural’ and ‘additive-free’ in Natural American Spirit (NAS) advertisements.1 In January 2017, SFNTC and the FDA reached an agreement under which NAS advertisements will no longer contain these words. Yet ‘natural’ will remain in the brand’s name, and related terms such as ‘organic’ and ‘tobacco and water’ remain allowable in the brand’s advertising, raising concern among public health advocates due to likely implications of reduced harm.2 3 The current study seeks to establish causal claims regarding the potential for exposure to NAS advertisements and the claims therein to produce inaccurate beliefs about product risk that invite product use.
Ample research suggests that despite requisite warning labels,4 descriptions on cigarette packaging that promote a ‘natural’ image foster perceptions of reduced risk for disease, especially among adolescents.5–9 Content analytic findings pertaining to NAS advertisements have identified several attributes, including textual claims about organic tobacco and sustainable growing practices, as well as images of plants, that could contribute to misperceptions of reduced harm.3 Focus group findings have demonstrated that smokers often believe NAS cigarettes pose lower health risks, doubt the veracity of pack warning labels and cite these misinformed beliefs as a reason for smoking NAS.10 Survey results further indicate that smokers, especially those who use NAS, believe the brand to be less harmful and attend to health considerations more than other smokers.11 12 While some studies have experimentally assessed the effects of NAS pack descriptors, the present body of research about NAS advertising, which potentially reaches a broader audience than the packs themselves, lacks sufficient experimental evidence regarding misinformation. To the authors’ knowledge, no study has yet established a causal path from exposure to NAS advertising to misinformed beliefs that influence intentions towards NAS.
To fill this gap, this randomised experiment was designed to test the effects of existing NAS advertising content on potentially inaccurate beliefs about NAS and their relevance to attitudes and intentions towards NAS. The goals were to (1) assess whether NAS advertisements and the claims used therein produce incorrect beliefs about NAS among current and former smokers; (2) identify types of beliefs that shape attitudes towards the brand; and (3) examine the relationships between exposure to NAS advertising content and attitudes/intentions towards the product. We hypothesised that exposure to NAS advertisements and advertising claims would produce misinformed beliefs about the make-up of NAS cigarettes and their health/safety, and that exposure would influence outcomes such as attitudes and intentions via belief change.
Methods
Participants
Participants for this study were recruited from an online survey panel provider (Qualtrics). All participants were age 18 or older, and either current smokers (ie, smoked at least 100 cigarettes in their lifetime and now smoke daily or some days) or former smokers (ie, previous daily smokers who quit at least 6 months prior).13 The final sample consisted of 1128 adults, including 412 daily, 238 intermittent and 478 former smokers. Of current smokers, 31% reported having ever tried NAS, and 5% selected it as their regular/preferred brand. Of former smokers, 11% had tried NAS, and 2% identified it as their (prior) regular/preferred brand. Respondent demographics by condition are presented in table 1.
Stimuli
To select the stimuli for the treatment conditions, we conducted a detailed review of NAS print and online advertisements available via search engines (eg, Google), the NAS website and/or organisational sites that compile tobacco advertisements (eg, the Stanford University Research into the impact of tobacco advertising). We categorised the advertisements into four treatment conditions (see online supplementary table 1): graphics-only advertisements (condition 1), simple advertisements (condition 2: imagery and embedded text), elaborated advertisements (condition 3: imagery and longer, embedded text) and NAS website advertisements (condition 4: imagery and accompanying text, from the NAS website). We identified prevalent themes and recurring claims in the advertisements, which we used to generate condition 5, a text-based ‘aggregated advertising claims’ condition that comprised arguments taken nearly verbatim from several NAS advertisements. The purpose of this condition was to more broadly represent the universe of NAS advertising claims and the cumulative effects of exposure to such claims. We expected that brief exposure to single advertisements might not sufficiently represent the spectrum of NAS marketing claims to which individuals are exposed in the real world. Because we are interested in the totality of effects of NAS marketing, we deployed a more comprehensive set of arguments taken directly from multiple advertisements.
Supplementary file 1
Conditions 1 and 2 each had three stimuli, whereas conditions 3, 4 and 5 each had four, for a total of 18 unique stimuli. Each advertisement (or textual compilation) represented one of several themes we identified in our search of NAS marketing (eg, natural, organic, taste themes). We differentiated the advertisements in each condition to promote sufficient representation of the thematic and presentational diversity of NAS advertisements. Still, we did not seek to test hypotheses regarding the effectiveness of different advertisement features, but rather to broadly ascertain the effects of exposure to NAS advertisements and the arguments therein.
Procedures
This study employed an online randomised experimental design with five treatment conditions and a no-message control group. All participants first completed background questions regarding demographic and smoking-related information. Eligible respondents were then randomised to condition (randomisation was successful). After assignment, control group participants completed the dependent measures, whereas message condition participants were exposed to existing NAS advertisements (conditions 1–4) or aggregated claims from the advertisements (condition 5), for a minimum of 5 s. Participants viewed two advertisements or text compilations randomly selected from the pool of materials in their assigned condition, and then completed dependent measures. All participants had to correctly answer two attention checks in the questionnaire in order to complete the survey. At the end of the survey, all participants were debriefed as to the harms of NAS and tobacco use.
Measures
Dependent measures included assessments of participant beliefs, attitudes and intentions regarding NAS. To inform the development of belief statements, we conducted an in-depth review of language in the NAS advertisements. Furthermore, from October 2015 to January 2016, research assistants conducted exhaustive searches of content (eg, news articles, discussion forums, videos) related to NAS on Google and YouTube, yielding a database of 295 articles and links to online discussions and 50 YouTube videos. We examined each entry, including the top 10 comments, and recorded all unique categories of beliefs as well as prototypical comments related to the categories that emerged. We then generated Likert-type belief statements pertaining to five themes arising from claims in NAS advertisements and from the public commentary about NAS. The thematic categories were (1) greater health/safety (reduced harm) of NAS cigarettes compared with traditional cigarettes; (2) organic/chemical-free (healthier) composition of NAS cigarettes; (3) lower environmental impact of NAS due to the company’s sustainable, US-based growing practices; (4) superior taste of NAS due to the natural composition; and (5) lower addictiveness of NAS due to the natural composition. Some statements addressed factual information (eg, whether all NAS cigarettes contain organic tobacco), while others addressed relative claims (eg, whether NAS cigarettes are a more responsible health choice than traditional cigarettes). The six attitude measures used slider scales representing semantic differentials about NAS attributes (eg, negative–positive), and we combined them into a single attitude scale (see online supplementary table 2). Intention measures were modified Likert responses pertaining to intentions towards NAS and traditional cigarettes (see online supplementary table 3). Measures are available from the authors upon request.
Data analysis
To address the first research goal of determining the effects of exposure to NAS advertisements and advertising claims, one-way analysis of variance compared condition means with control group means on beliefs, attitudes and intentions. Belief items were grouped into scales based on the five aforementioned themes: health/safety (alpha=0.90), chemical composition (alpha=0.68), sustainability (alpha=0.74), taste (alpha=0.87) and addictiveness (alpha=0.83). Attitude items were also scaled (alpha=0.94), as were the three NAS intention measures (alpha=0.89). The item pertaining to intentions to smoke traditional cigarettes was analysed individually.
To assess the plausibility of the relationships specified under research goals 2 and 3, a regression analysis of attitudes on beliefs tested which beliefs predict attitudes, and mediation analysis tested the relationships among message condition, beliefs, attitudes and intentions towards NAS.
Results
Results discussed below present the main effects of exposure on beliefs, attitudes and intentions, as well as regression and mediation analyses. Analyses are grouped by former or current (daily and intermittent) smoking status, as the interaction of daily smoking and condition was not significant for any belief scales.
Beliefs, attitudes and intentions
Tables 2 and 3 present the main effects of exposure on beliefs, attitudes and intentions for former and current smokers, respectively. Exposure to all message conditions resulted in more favourable ratings of NAS (relative to control) on one or more of the belief scales. In particular, current and former smokers in most experimental conditions reported more positive beliefs about the sustainability of NAS cigarette production and the chemical composition of NAS cigarettes, and the latter category represented several inaccurate statements (eg, that all NAS cigarettes are organic). Participants exposed to aggregated NAS advertising claims reported significantly more favourable beliefs towards NAS across all belief scales among former and current smokers. In all cases, significant differences from control represented more positive beliefs about NAS among those exposed to experimental stimuli. Furthermore, many of the beliefs reflected misinformation about NAS, particularly those related to health/safety, addictiveness and chemical composition (eg, cigarettes with organic tobacco are less harmful than traditional cigarettes) (see online supplementary table 4). In addition, participants who had tried NAS reported more favourable beliefs about NAS for most belief scales, including the inaccurate health/safety beliefs (see online supplementary table 5). Because not all conditions had warning labels, we tested whether the absence of a label led to more favourable beliefs about NAS; planned contrasts between conditions 1, 2 and 3 (with labels) and condition 4 (without) revealed no significant differences.
Current and former smokers exposed to the aggregated advertising claims condition also reported more favourable attitudes towards NAS than control group participants. Former smokers exposed to the condition using graphic NAS advertisements and current and former smokers exposed to aggregated advertising claims reported higher intentions to use NAS than control group participants, yet current smokers viewing aggregated advertising claims reported lower intentions to smoke traditional cigarettes than control group respondents.
Regression and mediation analyses
Multiple regression analysis (table 4) pooled across all conditions highlighted the salience of inaccurate health/safety beliefs. Results indicated that three belief scales explained 56% of the variance in former smokers’ attitudes towards NAS (F(4, 473)=149.95, p<0.001), with health/safety (β=0.56), taste (β=0.16) and environmental sustainability beliefs (β=0.13) significantly and positively predicting attitudes towards NAS. These predictors also explained 58% of the variance in current smokers’ attitudes (F(4, 645)=224.66, p<0.001): health/safety (β=0.48), taste (β=0.27) and environmental sustainability (β=0.13).
Mediation analysis in SPSS using Hayes’ PROCESS method14 15 indicated that each individual belief scale significantly, positively predicted attitudes towards NAS, which predicted current and former smokers’ intentions to smoke NAS. Mediation analysis further tested a pathway from message exposure to health/safety beliefs (as these beliefs accounted for the largest percentage of variance in attitude prediction) to attitudes and intentions. The mediation tests represented the a priori hypothesis that exposure to NAS advertisements and the claims therein would produce misinformed beliefs about the health/safety of NAS, which would in turn influence attitudes and intentions. Figures 1 and 2 present the mediation path from the aggregated advertising claims condition (vs the no-exposure control group), as this condition most consistently differed from control on belief scale means. Results support the expected path of influence from the aggregated advertising claims condition to attitudes to intentions, mediated through health/safety beliefs. For current and former smokers, significant results were obtained for the test of indirect effects from condition exposure to beliefs, to attitudes and then to intentions, whereas direct effects were not significant. As the observed indirect effect did not fully account for the total effect, results indicate that misinformed health/safety beliefs partially mediate the effects of condition exposure on intentions to use NAS.
Discussion
This study employed a rigorous experimental design to establish causal claims regarding the effects of exposure to NAS advertising content. To assess causal effects, our study compared beliefs of participants randomly assigned to exposure to NAS advertisements or aggregated advertising claims versus a control group with no exposure. Prior research applied content analysis, focus groups and survey methodology to identify misleading claims in NAS advertising that may prompt incorrect beliefs and influence intentions, but this study applied a randomised experiment to test such assertions. Our study found that exposure to NAS advertisements or aggregated arguments from the advertisements respectively produced inaccurate beliefs about the composition and health/safety of NAS cigarettes, demonstrating that the range of suggestive arguments in NAS marketing yields spillover effects onto inaccurate beliefs not directly addressed in the advertisements. The experiment further showed that inaccurate health-related beliefs heavily predict current and former smokers’ attitudes towards NAS, and mediate intentions to use the product.
The treatment conditions represented a snapshot of the universe of existing, publicly available NAS advertisements and the claims embodied therein at the time of data collection. All message conditions yielded more positive beliefs about NAS relative to control in at least one belief area that gives rise to misperceptions about the product (ie, chemical composition or health/safety). Specifically, all conditions produced more favourable, inaccurate beliefs regarding the composition of NAS, so even brief exposure to NAS advertisements prompted the generation of composition-related misinformation. Still, the beliefs, attitudes and intentions of participants in the aggregated advertising claims condition differed most consistently from control. Perhaps this is because brief exposure to specific arguments in just two native NAS advertisements cannot fully capture the real-world effects of the wide variety of NAS advertisements. Yet the aggregated advertising claims condition more closely represented the spectrum of appeals in NAS marketing because it was an aggregation of claims taken directly from numerous NAS advertisements. This condition may have yielded more consistent effects on misinformed beliefs, and on attitudes and intentions, because it was designed to holistically encapsulate the broad arsenal of arguments in NAS advertisements; the total effects of NAS marketing on the public are not restricted to exposure to arguments from a single advertisement. Importantly, exposure to these aggregated arguments produced inaccurate beliefs about reduced harm and improved healthfulness of NAS cigarettes (eg, NAS cigarettes are safer than traditional cigarettes), even though the advertisements do not make explicit health-related claims. The ‘effectiveness’ of this condition in generating more favourable, and in many cases incorrect, beliefs about NAS demonstrates potentially harmful aggregate effects of the varied claims used by SFNTC to establish the NAS image.
Mediation and regression analyses confirmed paths of influence from beliefs to downstream outcomes such as attitudes and intentions towards trying the product. While the mediation analyses conducted are not definitive in establishing a causal sequence from beliefs to attitudes, the posited sequence we tested is consistent with a long established causal order from behavioural change theories.16 17 The significance of our mediation results is twofold. First, they indicate that all belief categories, including those representing factually inaccurate information, predict attitudes and intentions. Perhaps more significantly, the mediation and regression results indicate that health/safety beliefs are particularly valuable in predicting attitudes towards NAS. These results signify that explicit (and ostensibly accurate) claims about the sustainability of NAS cigarettes, for instance, are less influential in motivating attitudes and intentions to use NAS than are misguided assumptions about health/safety implied by the arguments in NAS advertisements.
The effect of the experimental conditions is particularly meaningful because the two groups of participants in this study—current and former smokers—should not be especially predisposed to unreasonably favourable attitudes regarding NAS in particular. Although the majority of current smokers in the study had never tried NAS, analysis revealed more positive attitudes among those exposed to message treatments (advertisements and/or arguments), based on statements about NAS and on comparisons to traditional brands. Relative to control group participants who had not seen NAS advertisements or advertising claims, those who had were more likely to denigrate their own brand (ie, ‘traditional cigarettes’) in comparison to NAS. This trend offers a potential explanation for the fact that current smokers in the aggregated claims condition reported lower intentions to smoke traditional cigarettes, while simultaneously exhibiting greater intentions to use NAS. Alarmingly, the results also illustrated more positive beliefs, attitudes and intentions regarding NAS and its safety among former smokers—individuals who undertook the difficult task of cessation potentially because of smoking’s harmful effects.
Limitations of this research pertain to the sample used and the constraints on exposure. The sample here differed primarily from a nationally representative smoker sample in that it had a larger percentage of white participants (87% compared with 66%) and a larger proportion of respondents reporting some college or more (50% compared with 42%).18 Still, these divergences make the sample more typical of the target population of NAS smokers, as the 2013 Population Assessment of Tobacco and Health study illustrated that the NAS smoking population tends to skew to more educated (75% weighted average with at least some college) and more white (88% weighted average white) than the general population of smokers.12
Another potential limitation of this study pertains to the nature of exposure in the message manipulations. The aggregated claims condition represented messages that participants would not have previously encountered in that exact form, and this novelty may have contributed to their effectiveness. Further, unlike the other conditions, this condition did not contain visuals; still, this divergence should not have driven the observed ‘effectiveness’ of the condition, as research indicates that visuals tend to be more effective. Conversely, participants may have already been exposed to the advertisements in the other conditions, and this familiarity could have muted the ability of the experiment to detect the full extent of their real-world effects. Viewing just two potentially familiar advertisements for a few seconds might not adequately represent the cumulative effects of exposure to multiple NAS messages in the real world; this may explain why the aggregated advertising claims condition was more impactful in terms of beliefs. The limited exposure in the experiment makes any observed effects conservative in their estimated size.
Conclusions
This study experimentally assessed the issue of whether NAS advertisements and the range of arguments presented therein affect beliefs about NAS (including inaccurate beliefs about product health/safety) and whether such beliefs influence attitudes and intentions regarding NAS. Results indicate that the answer to those queries is yes. Our findings offer experimental evidence supporting implications from previous non-experimental research that the image cultivated through NAS advertising, including the focus on being natural, organic and US-grown, may have direct implications for individuals’ beliefs about the product’s composition, safety and healthfulness. Further, our results directly establish the salience of inaccurate health/safety beliefs by empirically linking them to attitudes and intentions towards NAS. These findings are especially concerning due to the growing popularity of NAS,19 20 and evidence that the disclaimers required thus far does not sufficiently dispel consumer beliefs that NAS poses fewer health risks than other cigarettes and, more generally, that health disclaimers are ineffective in countering misperceptions about product safety.10 21–23 They also raise concern considering the limited scope of the agreement between SFNTC and the FDA.2
Prior studies have highlighted that no conclusive evidence indicates that the advertised ‘natural’ attributes of NAS make NAS cigarettes any less harmful than traditional cigarettes; thus, as research continues to provide validation of prior non-experimental findings about NAS advertising effects (which our study does), ‘it may be appropriate for regulatory agencies to ban ‘natural’, ‘organic’ and ‘additive-free’ descriptors’ (p520), a regulatory action which such agencies have not yet taken.10 As a result, the findings of this study offer further evidence for regulators considering the ramifications of claims used in the marketing of natural cigarettes. To build on the findings here, recommended next steps would be to assess implicit attitudes towards NAS brought about by exposure to their advertisements, and to test message designs aimed to target implicit and explicit attitudes and to remedy potentially inaccurate beliefs about NAS.
What this paper adds
Using content analysis, focus groups and survey methodology, prior research identified potentially misleading claims in Natural American Spirit (NAS) advertising that may affect beliefs about the product and decisions to use it. This study applied a randomised experimental design to test the effects of exposure to NAS advertising content, and potential mechanisms of that effect, among adult current and former smokers.
Results demonstrate causal effects of NAS advertisements and the variety of arguments contained therein on inaccurate beliefs about the health, safety and composition of NAS cigarettes and attitudes towards NAS; findings indicate that exposure to these advertisements/arguments cues intentions to use the product even among former smokers.
This research established pathways of influence from beliefs about NAS (prompted by exposure) to intentions, by way of mediating attitudes. Inaccurate beliefs about reduced harm and improved healthfulness driven by advertising content were shown to be particularly strong predictors of attitudes towards NAS and to mediate effects on intentions.
Acknowledgments
The authors would like to thank Kyle Cassidy, digital design specialist at the Annenberg School for Communication, for his assistance in preparing the experimental stimuli and figures, as well as Yotam Ophir, PhD candidate at the Annenberg School for Communication, for his assistance in the programming of this study.
Footnotes
Contributors All four authors met the tobacco control criteria for authorship and can serve as guarantors of the paper. All authors participated in the design and administration of the study, as well as the writing and revising of the paper. SKG is the lead author.
Funding Data collection was supported by the National Cancer Institute (NCI) of the National Institutes of Health (NIH) and the Center for Tobacco Products (CTP) of the Food and Drug Administration (FDA), under Award Number P50CA179546. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the FDA.
Competing interests None declared.
Ethics approval All procedures were approved by the University of Pennsylvania Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement For this study, we also collected data on participant evaluations of the advertisements they viewed. Additional data may be available upon request from the authors and with approval of the University of Pennsylvania Tobacco Center on Regulatory Science.