Article Text
PDF
Abstract
Objective To examine which types of cessation-focused advertisements are associated with perceived advertisement effectiveness among smokers and to assess whether key smoker characteristics (ie, cigarette consumption, desire to quit and past quit attempts) influence perceived effectiveness of different types of cessation ads.
Methods We used data from the New York Media Tracking Survey Online, a web survey of 7060 adult smokers in New York. Participants were exposed to four categories of cessation ads: (1) why to quit—graphic images, (2) why to quit—testimonial, (3) how to quit and (4) anti-industry. Perceived ad effectiveness was measured with a four-item scale assessing the degree to which participants thought the ads made them stop and think, grabbed their attention, were believable and made them want to quit smoking. We categorised smokers based on cigarette consumption, desire to quit and past quit attempts. We used multivariable analyses to examine how smoker characteristics and category of cessation ads predict perceived ad effectiveness.
Results Ads using the ‘why to quit’ strategy with either graphic images or personal testimonials were perceived as more effective than the other ad categories. Smokers who had less desire to quit or had not tried quitting in the past 12 months responded significantly less favourably to all types of cessation ads tested. Greater cigarette consumption was also associated with lower perceived effectiveness, but this association was smaller in magnitude.
Conclusions Tobacco control programmes that utilise cessation-focus advertising should focus relatively more on ads that adopt the ‘why to quit’ strategy with either graphic images or personal testimonials.
- Smoking cessation
- media
- social marketing
- advertising and promotion
- cessation
Statistics from Altmetric.com
Introduction
Mass media campaigns are a prominent tool in promoting smoking cessation and have generally been found to be effective in promoting cessation-related outcomes.1 2 Cessation-focused campaigns have employed a fairly wide variety of message themes. Among these, three of the most common broad themes for cessation campaigns include: (1) why to quit, (2) how to quit and (3) anti-tobacco industry.3 Although there is considerable variation in the specific execution of these broad themes, messages that focus on why to quit typically involve the use of graphic images and/or testimonials that evoke specific emotional responses that portray the serious health effects of smoking. Messages that focus on how to quit are generally informational in nature, providing smokers with support in the quitting process with websites, quitline numbers and ways to get started with quitting. Anti-tobacco industry messages focus on the questionable marketing practices of the tobacco industry. Other themes for cessation have been used more sparingly, including messages that focus on the social acceptability of smoking, the effects of smoking on physical appearance, and the perils of addiction. Although some antismoking messages also focus on the effects of secondhand smoke, these ads are often intended to promote support for clean indoor air policy and are generally less focused on cessation itself.3
Advertising effectiveness by message theme
Recent studies suggest televised messages that use graphic images to depict the negative health consequences of smoking4 and messages that use emotive testimonials depicting loss of family and other smoking-related hardship5–7 may be effective in promoting cessation and reinforcing smokers' intentions to quit.8 These findings are consistent with a more extensive literature on adolescent response to antismoking ads that shows ads with visceral, evocative imagery9 10 that illicit negative emotional responses11 are more readily recalled and perceived as more effective among adolescents. Parallel research on antidrug messages reaches similar conclusions about the role of emotional content (eg, Morgan et al and Donohew et al12 13).
Some evidence suggests that cessation messages that emphasise the benefits of quitting (ie, gain-framed messages) are efficacious.14–16 However, the implications of these findings for televised media campaigns are limited because this research focuses almost entirely on message argument and language (ie, gain versus loss-framing). Other emerging research suggests that characteristics of ads, such as stylistic features and message themes (eg, use of graphic images, emotional content, why-to-quit and how-to-quit themes), also have significant effects on message recall, persuasion and processing.9 17–19 Inattention to these details can compromise effective and persuasive message design, a key goal of communication theory.20 While existing research on framing and effects of ad characteristics is informative, little is known about how these effects may differ by smokers' cigarette consumption, desire to quit or prior cessation experience.
Advertising effectiveness by smoker characteristics
Consistent with communication theories such as the elaboration likelihood model (ELM),21 an individual's smoking history and motivation to quit may influence message processing. It is argued that for a message to be persuasive, the individual viewing the message must attend to and process its content. This may occur through central processing routes that involve effortful consideration of the message or peripheral routes that involve less intensive information processing and more immediate reactions to messages.22 Smokers may differ significantly in terms of message processing. For example, smokers interested in quitting or who have previously tried quitting may be more likely to centrally process cessation messages. Among smokers that are less motivated to quit, message processing routes could be dependent on both how concerned the smoker is about quitting and how willing the smoker is to quit. For example, smokers who are simply unconcerned about quitting (eg, young smokers who may discount long-term effects while they are young) may be unmotivated to process messages and thus process them peripherally when they are exposed to them. Conversely, smokers who are simply unwilling to quit (eg, older smokers who may simply enjoy smoking and are defiant in their smoking behaviour) may be motivated to process cessation ads centrally and, as a result, actively counter-argue those messages. If processing of cessation ads differs significantly by smoker characteristics in this way, then improved knowledge of how different types of smokers respond to varying cessation messages can greatly inform the development of future cessation messages.
Perceived effectiveness predicts actual effectiveness
Experimental studies in health communication suggest that perceived effectiveness of ads, which involves cognitive and emotional responses, predicts changes in attitudes and other behavioural precursors related to the outcomes targeted by messages. For example, Dillard and colleagues23 24 showed that measures of ‘perceived persuasiveness’ of public service announcements (PSAs) predicted changes in attitudes towards the social issues and subject matter of the PSAs. This research is consistent with guidance provided by the ELM,21 which posits that persuasion is a product of the mental processes evoked by a message. As such, measures of reactions to ads can be valuable upstream indicators of message effectiveness.
While prior literature has established a relation between perceived effectiveness and actual ad effectiveness, existing measures of perceived effectiveness are somewhat one-dimensional in that they only capture general emotional responses, such as surprise, anger and fear.24 Although these are important reactions there may be other critical responses specific to the content of the ads themselves. Furthermore, measures that focus on emotional response may be biased towards ads that rely on emotional evocation to convey their central message, reducing the validity of claims based on these measures. An ad focused on how to quit may not evoke emotional responses but could be quite effective in changing behaviour and promoting cessation. To our knowledge, no studies have used reliable item scales for assessing smokers' perceived effectiveness of cessation ads. Reliable measures of reactions to ads that are specific to message content can enhance campaign designers' ability to evaluate cessation message strategies and improve media planning.
In this study, we examine data from an online survey of smokers in New York State to test the reliability of a new perceived effectiveness scale for smoking cessation advertisements. We first conduct principal factor analyses on survey items that measure specific smoker reactions to a number of cessation ads to determine a reliable scale measure of perceived effectiveness. We then perform descriptive and multivariable analyses on this measure to address two primary research questions: (1) which styles of cessation ads predict the highest perceived effectiveness among smokers overall? and (2) are there key smoker characteristics (cigarette consumption, interest in quitting, and quit attempts) that predict perceived effectiveness of cessation ads? Implications of our findings for current and future development of cessation-focused media messages are discussed.
Methods
Survey data
Our data come from the New York Media Tracking Survey Online (MTSO), a self-administered online survey of adult smokers in New York State. The primary purpose of the MTSO is to track smokers' response to and awareness of specific antismoking television ads planned to air (summer surveys) or actually aired (spring surveys) by the New York State Department of Health. The MTSO sample was drawn from New York residents who participate in the Harris Poll Online. The survey was conducted in five waves over 2 years in spring 2007, summer 2007, spring 2008, summer 2008 and spring 2009. The combined dataset consisted of 7060 unique adult smokers in New York. A small subset of participants completed more than one survey wave, allowing for sufficient sample sizes across waves for testing of all ads. The final sample included 5785 white (81.9%), 371 African American (5.3%), 399 Hispanic (5.7%) and 505 other/unidentified (7.2%) adult smokers. The majority of the sample was female (64.5%). The MTSO instrument, consent procedures and human subjects protection protocols were reviewed and approved by the sanctioned institutional review boards of RTI International and the New York State Department of Health.
The MTSO data were weighted to make the sample reflect the demographic composition of New York smokers statewide. The MTSO weights adjust for attitudinal and behavioural differences between internet and non-internet households, differences by likelihood to join online panels and differences between MTSO respondents and non-respondents. Table 1 summarises weighted and unweighted MTSO sample sizes by gender, age, race and education as well as weighted demographic characteristics from the 2008 New York Adult Tobacco Survey (ATS), a statewide representative survey of smokers in New York State. The sample weight generally increases representation of racial/ethnic minorities, males, and those with lower educational attainment, increasing the similarity between the MTSO and ATS samples.
Media tracking survey online (MTSO) weighted and unweighted sample sizes (unique individuals) and comparable weighted demographics from the adult tobacco survey (ATS)
Cessation ad stimuli
A key feature of the MTSO was the ability to show videos of antismoking ads to participants via online multimedia tools. This feature, a validated method for measuring encoded exposure,25 also enabled us to measure smokers' reactions to specific ads immediately after viewing them rather than relying on longer-term recall. Each participant was exposed to videos of specific ads within the survey. Participants were shown ads from each of four categories: (1) why to quit—graphic images, (2) why to quit—testimonial, (3) how to quit and (4) anti-industry. These ad categories are described in detail below.
Ads were categorised independently by two trained coders. Ads illustrating the health effects of smoking using graphic images (eg, clogged arteries, black lung) were coded as ‘why to quit—graphic images’. This category included ads such as those from the ‘Every Cigarette Does You Damage’ series, which shows the effects of smoking on specific organs in the body. Ads that demonstrate pervasive negative emotions through personal testimonials (eg, death, loss of loved ones, physical disabilities) were coded as ‘why to quit—testimonial’. This included ads such as ‘Rick Stoddard’, which features a man talking about how his wife died at an early age from lung cancer. Ads focusing on the quitting process and support for quitting (eg, websites, quitlines) were coded as ‘how to quit’. These ads usually feature a specific quitting service with positive messages that are sympathetic to the difficulties smokers face in quitting. Finally, ‘anti-industry’ ads focused on the tobacco industry's marketing practices (eg, glamorous portrayals of smoking, attempts to hide dangers of smoking). Cohen's kappa coefficient was calculated after coding to measure inter-coder agreement. The kappa coefficient was 0.77, indicating substantial reliability. Discrepancies between the coders were discussed until consensus was reached.
In total, we assessed 37 cessation-focused ads, including 10 ‘why to quit—graphic images’ ads, 15 ‘why to quit—testimonial’ ads, 8 ‘how to quit’ ads and 4 ‘anti-industry’ ads. Because our data come from the MTSO, which served the primary function of tracking awareness of ads that were actually aired in New York, the number of ads included in each theme category is a direct function of the state's campaign implementation and not a pre-determined study design feature. The exact number of ads viewed by each respondent varied across each MTSO wave, and ranged from four to six ads depending on how many ads were being tested for planning purposes or being aired in New York and available to track. To avoid significant increases (and related costs) in sample sizes during waves of heavy advertising, the limit of exposure was increased to six ads. High-resolution videos were shown to participants with broadband internet connections, whereas low-resolution versions were shown to participants with dial-up connections. All ads included in this study were 30 seconds in length and the order in which ads were presented to any single participant was randomised.
Person-ad data
The data were structured to contain unique observations for each viewing of an ad. For example, an individual who viewed four ads would correspond to four distinct observations for the four ads that person viewed. This person-ad structure permits analysis of perceived effectiveness and other ad-specific measures for all ads overall and separately by each style of ad. The resulting dataset contained a total of 33 447 person-ad observations. The presence of repeated observations on the same individual implies that perceived effectiveness may be correlated across all ads seen by the same individual. We thus adjusted all estimated standard errors by clustering all analyses on a unique individual identifier.
Survey measures
Perceived ad effectiveness
Perceived effectiveness was measured separately for each ad, immediately after the ad was viewed. Participants were asked to provide feedback on each ad by answering questions that measure aspects of ad persuasiveness, believability and processing. Two items asked how much the ad made them stop and think and how much the ad grabbed their attention using a Likert response scale of 1 (strongly disagree) to 4 (strongly agree). Respondents were also asked to indicate the degree to which they found the ads believable on a scale of 1 (‘not at all’) to 5 (‘very’). Finally, respondents were asked to indicate, on a scale of 1 to 5, how much the ad made them want to quit smoking. For each item, higher response values indicated greater perceived effectiveness. We created a perceived effectiveness scale equal to the linear sum of the items. This measure is similar to other constructs (eg, Dillard et al and Bruner23 26) that assess cognitive reactions and attitudes towards ads. While there were two different response scales (1 to 4 and 1 to 5) across these four items, we chose to not standardise the item scores to z-scores before summing them into the perceived effectiveness scale. This was done for two reasons. First, the unstandardised perceived effectiveness scale is more interpretable in the descriptive analyses presented below. Second, we replicated all analyses in this study using a standardised version of the perceived effectiveness scale and found no differences in the qualitative results and conclusions. Therefore, we use the simpler linear sum perceived effectiveness scale.
Cigarette consumption, interest in quitting and quit attempts
We analysed perceived effectiveness of cessation ads by three specific smoker characteristics: (1) cigarette consumption, (2) desire to quit and (3) past cessation attempts. Cigarette consumption was measured with a categorical variable for daily cigarette consumption (<10, 10–19, and 20+ cigarettes per day). Desire to quit was based a survey item asking smokers ‘how much do you want to quit smoking’ with responses for ‘not at all’, ‘a little’, ‘somewhat’ and ‘a lot’. Our construct for desire to quit was measured with an indicator variable for whether the smoker currently wants to quit ‘a lot’. The MTSO asked smokers whether during the past 12 months they have stopped smoking for one day or longer because they were trying to quit. Smokers who indicated ‘yes’ to this question were considered to have made a quit attempt in the past 12 months.
Individual characteristics
A number of individual characteristics were measured and included as control variables in our models for perceived effectiveness of cessation ads. These included gender, age, race/ethnicity, education and income. Age was measured with separate indicator variables for ages 18–24, 25–39, 40–64 and 65 or older. Race/ethnicity was measured with indicator variables for white, African American, Hispanic and other races. Education was measured using indicator variables for having less than some high school, some high school education, high school graduate only, having some college education and having at least a college degree. Income was measured with indicator variables denoting intervals of $14 999 or less; $15 000 to $24 999; $25 000 to $34 999; $35 000 to $49 999; $50 000 to $74 999; $75 000 to $99 999; and $100 000 or more in annual income. Because many ads previously aired in New York we also included an indicator variable for whether respondents had previously seen the ad they viewed. Our models also included an indicator variable for whether the participant had high speed 9nternet to control for quality of the video stimuli.
Statistical analysis
Perceived effectiveness scale reliability and validity
We conducted principal factor analysis to determine whether the perceived effectiveness items form a single scale.27 This was accomplished with the principal factor method in Stata statistical software and was done separately for each ad category. We examined eigenvalues for identified factors, loadings for retained factors and proportion of variance for retained factors to verify the single factor solution. Scale reliability was then assessed by estimating Cronbach's α coefficient overall and separately for each ad category. To assess scale nomological validity, we examined associations between perceived effectiveness and beliefs related to the message content. We expected those who indicate a higher perceived effectiveness of an ad to be more likely to hold beliefs consistent with that ad's content. This expectation was based on the belief that perceived effectiveness should be positively correlated with a measure of actual effectiveness.23 We recognise that a cross-sectional association between perceived effectiveness and message-related beliefs does not provide strong evidence of actual effectiveness; the temporal order between these variables has not been established. Nevertheless, positive associations are considered a necessary (but not sufficient) condition for the possibility of a causal relation and thus provide a reasonable (albeit limited) criterion with which to assess nomological validity.
Descriptive and multivariable analysis of perceived ad effectiveness
Descriptive and multivariable analyses were used to gauge smokers' perceived effectiveness of ad categories. We assessed average perceived effectiveness of ads from each of the four main ad categories among smokers overall and by each smoker characteristic we measured. We then estimated ordinary least squares models to assess the relation between the perceived effectiveness scale, ad category and each smoker characteristic described above. Perceived effectiveness was modelled as a function of cigarette consumption, interest in quitting and past quit attempts, controlling for gender, age, race\ethnicity, education, income and prior ad exposure. We also included indicator variables for ad category (why to quit—testimonial, how to quit, and anti-industry) with why to quit—graphic ads as the reference group. Separate models were also estimated for perceived effectiveness of each of the four ad categories.
Results
Principal factor analyses of the four perceived effectiveness variables retained only one factor, with an eigenvalue of 2.51 and factor loadings ranging from 0.72 to 0.87 across the four variables (table 2). The proportion of total variance accounted for by this single factor was 0.63 and reliability of the single four-item scale was high (Cronbach's α 0.87). Results were similar within each ad category. The scale was also distributed fairly normally across all ads, with a mean perceived effectiveness of 12.5 (SD 3.77). We found some limited evidence of validity for the scale. The correlation between perceived effectiveness of the ‘Gangrene’ ad and agreement (measured as ‘strongly agree or agree’) with the statement that ‘smoking causes infections like gangrene’ was substantially higher than the correlation between perceived effectiveness of any other ad (or across all ads) and that specific belief statement. This suggests that this measure may capture specific message processing linked to individual ad content.
Descriptive statistics and reliability analysis of the perceived effectiveness scale for all ads
In our person-ad dataset, 32.9% of smokers indicated wanting to quit a lot and 47.3% of smokers reported having made at least one quit attempt in the past year. In terms of cigarette consumption on days smoked, 29.2% of smokers indicated smoking less than 10 cigarettes per day, 31.7% indicated smoking between 10 and 19 cigarettes per day and 39.1% reported smoking 20 or more cigarettes per day.
Overall perceived effectiveness among all smokers was 13.1 (SE 0.10) for ‘why to quit—graphic images’ ads (table 3) and was significantly greater than all other ad categories (p<0.05). The ‘why to quit—testimonial’ ad category received the next highest average score of 12.8 (SE 0.09), statistically higher than ads in either the ‘how to quit’ (11.9 average perceived effectiveness, SE 0.10) or ‘anti-industry’ (12.0 average perceived effectiveness, SE 0.11) categories (p<0.05). In terms of standardised mean difference statistics (Cohen's d), the ‘effect size’ of the difference between ‘why to quit—graphic images’ and ‘anti-industry’ ads is approximately 0.2.28 Although some consider effect sizes of this magnitude to be ‘small’, this value is larger than the average effect size associated with antismoking media campaigns (r = 0.05, corresponding to d = 0.10), which are widely considered effective in reducing population smoking rates.2 29
Average perceived effectiveness by smoker characteristics and cessation ad categories (95% CIs)
Perceived effectiveness also differed by individual smoker characteristics. Cessation ads in the ‘why to quit—graphic images’ category received an average perceived effectiveness of 14.6 (SE 0.18) from smokers who want to quit a lot compared with 12.5 (SE 0.13) among smokers who do not want to quit a lot. Similar differences were observed by desire to quit for each of the other ad categories. These differences were statistically significant (p<0.01). Past quit attempts also differentiated smokers in terms of perceived effectiveness. Ads in the ‘why to quit—testimonial’ category were given an average perceived effectiveness of 14.0 (SE 0.11) from smokers who made a quit attempt in the past 12 months compared with 12.3 (SE 0.10) from smokers who did not make a past-year quit attempt. Similar differences were observed for all categories of ads and were also statistically significant (p<0.01). We also found that perceived effectiveness differed based on cigarette consumption, although the magnitude of these differences was smaller and limited to ads in each of the ‘why to quit’ categories. Specifically, ads in the ‘why to quit—graphic images’ category were given an average perceived effectiveness of 13.7 (SE 0.21) from smokers who smoke less than 10 cigarettes per day compared with 13.1 (SE 0.24) from smokers who smoke 20 or more cigarettes per day (p<0.05). Similarly, smokers who smoke less than 10 cigarettes per day reported an average perceived effectiveness of 13.4 (SE 0.12) for ads in the ‘why to quit—testimonial’ category compared to an average perceived effectiveness of 12.7 (SE 0.17) given by smokers who smoke 20 or more cigarettes per day (p<0.01).
Results from our ordinary least squares regression models support the pattern of findings in descriptive analyses (table 4). ‘Why to quit—graphic images’ ads were associated with significantly greater perceived effectiveness compared with all other categories of ads (p<0.01) after controlling for all other confounding variables in our model. We found that smoking 20 or more cigarettes per day was associated with lower perceived effectiveness overall and for the ‘why to quit—testimonial’ ads specifically. We did not find associations between cigarette consumption and perceived effectiveness for any other specific ad category. Low interest in quitting was associated with lower perceived effectiveness of all categories of ads (p<0.001). Absence of past 12-month quit attempts was also associated with lower perceived effectiveness of all categories of cessation ads. Overall, desire to quit was the strongest predictor of perceived effectiveness of cessation ads in terms of coefficient magnitude.
Multivariable linear regressions showing relation between smoker characteristics, ad categories and perceived effectiveness of cessation ads (95% CI)
Discussion
This study demonstrates a new four-item scale that appears to measure a single construct of perceived ad effectiveness. This scale is interpreted as a measure of how well a specific cessation advertisement resonates with a given smoker. We believe the items that constitute this scale capture cognitive processing as outlined by the ELM whereby individuals who respond more favourably to specific message content are more likely to be persuaded by the ad. We found the scale to have high reliability and although our ability to directly assess construct validity was limited, we did find some indirect evidence of validity. Based on these results, we deemed the scale a reasonable indicator of perceived effectiveness and examined smoker and ad characteristics that influenced these perceptions.
Consistent with prior literature,4 findings suggest that ads that use graphic images to depict the physical consequences of smoking are perceived as most effective among smokers. While the ‘why to quit – testimonial’ approach was rated slightly lower for perceived effectiveness, both of the ‘why to quit’ approaches were rated significantly more favourably compared to the ‘how to quit’ and ‘anti-industry’ classes of ads. This difference was apparent for all smokers overall and for each subgroup of specific smoker characteristics. One potential conclusion from these findings is that for all smokers, it may be useful to use the ‘why to quit’ strategies (with graphic images and emotive content) more often than ‘how to quit’ and ‘anti-industry’ strategies. However, this conclusion should be made with some caution as the ‘how to quit’ and ‘anti-industry’ ads included in this study certainly do not represent all possible ‘how to quit’ and ‘anti-industry’ strategies. It is quite possible that a differently executed ‘how to quit’ or ‘anti-industry’ strategy could yield higher levels of perceived effectiveness.
We also find that desire to quit and prior attempts at quitting are strongly associated with perceived effectiveness of cessation ads. Smokers who have less desire to quit or have not tried quitting in the past 12 months respond significantly less favourably to all types of cessation ads tested in this study. Cigarette consumption (smoking 20 or more cigarettes per day) was also associated with lower perceived effectiveness but this association was relatively smaller in magnitude and appears to be limited to ‘why to quit—testimonial’ ads. Given prior experimental research30 that links perceived effectiveness to behavioural precursors, this suggests that cessation ads we tested may be less effective among smokers who have lower desire to quit, have not tried quitting before and smoke a pack or more cigarettes per day. This is a potentially important consideration for campaign planning and development, particularly since the ads we tested are fairly representative of those used in state-funded campaigns in the USA and around the world. Quick interpretation of these findings might imply that a more customised approach to developing ads targeted specifically to these types of smokers would be useful. In practice, however, the implementation of such a recommendation may be limited for a number of reasons. First, it is quite possible that less motivated smokers will respond less favourably to any type of cessation-focused message regardless of how well it is targeted to their specific situation. Second, lower perceived effectiveness among this group of smokers may not necessarily translate into a negative impact on their smoking behaviour. Further data would be needed to explore this possibility. Third, state tobacco control programmes often do not have sufficient resources to develop wholly new messages and advertisements. These programmes most often adapt existing advertisements from the CDC's Media Campaign Resource Center, a repository of antismoking advertisements from the USA and other countries around the world, for their own state. Furthermore, the finding that less motivated smokers responded less favourably to all types of ad strategies examined in this study suggests that it may be extremely difficult to create effective ads that would change that pattern. Taken together, these limitations suggest there may be limited benefit in developing new cessation messages to appeal specifically to less motivated smokers.
Our results also highlight other research questions and issues to be addressed in future research. For example, many other specific ad characteristics besides major categories of theme and style may influence perceived effectiveness of cessation ads. A growing body of research suggests that specific and quantifiable stylistic features have meaningful implications for message attention, recall and processing.9 13 These specific ad characteristics may include the use of statistics and supporting evidence, ad style (eg, testimonial, dramatisation), number of edits and other features. Wong and colleagues31 examined the interaction between stylistic and production features of cessation ads, measured as ‘message sensation value’ (MSV) and found that MSV can intensify smokers appraisals of cessation ads that utilise death appeals, much like the ‘why to quit—graphic images’ ads examined in the current study. Similar work has been done and is ongoing in the realm of youth antismoking ads.9 10 Given the results presented in the current study, future research on this topic should further explore the interactions between MSV and each of the specific ad categories examined in this study.
Future research should also consider additional measures of perceived effectiveness itself. The items included in our scale are based on evaluative assessments of believability, attention-grabbing qualities and degree of engendering desire to quit. These qualities may be important in the assessment of all types of ads. However, there are multiple types of responses to ads that can vary substantially by ad type and design. For example, ads in the ‘why to quit—testimonial’ category are intended to generate a strong emotional response. Emotions represent a qualitatively different dimension of attitudes toward an ad, and prior research has found that emotions elicited by ads can have an impact on advertising effectiveness.32 While emotional evocation may be consistent with the purpose of ‘why to quit’ ads, it is unclear what kinds of emotional response promote eventual behaviour change. Thus, future research should seek improved understanding of individual emotions generated by cessation ads.
The findings in this study are subject to a few limitations. First, our findings may not be generalised to the broader population of smokers in the USA. The MTSO sample is restricted to New York smokers who may reflect a demographic composition different from the remainder of the country. New York smokers are also subject to one of the most stringent antitobacco policy environments in the USA, including the highest state cigarette excise taxes in the country, a myriad clean indoor air laws, and point-of-sale antismoking print advertisements in convenience stores. These factors may influence New York smokers' overall attitudes related to tobacco and their reactions to cessation-focused advertising. The MTSO is also an online survey that does not use probability-based sampling methods. Although the Harris Poll Online is weighted to statewide demographic benchmarks for New York residents, the opt-in sampling method of the panel limits our ability to generalise study findings. The MTSO is also limited in terms of measures that allow rigorous assessment of validity for our measure of perceived effectiveness. We were able to examine the association between perceived effectiveness of at least one ad (‘Gangrene’) and agreement with a belief statement that is directly tied to its content (‘smoking causes diseases such as gangrene’). However, the MTSO did not contain other belief measures that are as closely linked to specific ad content. Although these limitations are important to acknowledge, it should also be noted that two key strengths of the MTSO study is that it uses real smokes and includes real-world antismoking advertisements that have actually aired in the context of a statewide tobacco control programme.
Another potential limitation of the study is that we were not able to explicitly control or match ad production features across all ads seen by any one individual participant. Because the main function of the MTSO is to track antismoking ads that actually air in New York State, the mixture of ads that respondents were exposed to in a given MTSO wave was wholly subject to the campaign's current advertising profile at any given time. As such, the mixture of ads in terms of stylistic production features (eg, screen transitions, edits and cuts, music pace, use of statistics, etc) was not constant across all survey participants. Finally, our study is limited by cross-sectional analyses that do not specifically link perceived effectiveness to changes in cessation-related intentions and behaviours. Future research should use longitudinal data to investigate the relation between perceived effectiveness of cessation ads and downstream indicators of effectiveness.
In summary, this study suggests that ads that use graphic images to depict the physical consequences of smoking are perceived as most effective among smokers overall. In terms of individual smoker characteristics, we also find that desire to quit and past experience with quit attempts are strongly associated with perceived effectiveness of cessation ads while cigarette consumption has a more limited association. Thus, the types of cessation ads we examined may be less effective among smokers who have less desire to quit or have not tried quitting before. Media campaign planners should closely consider these issues in developing future cessation media messages.
What this paper adds
This study demonstrates a new four-item scale for measuring perceived ad effectiveness. We found this measure to have high reliability and it appears to be a reasonable indicator of how well a specific cessation advertisement resonates with a given smoker. This study also adds to a growing evidence base which shows that ads that use graphic images to depict the physical consequences of smoking and ads that use testimonials of personal loss from smoking are perceived as most effective among smokers overall. Furthermore, smokers clearly differ in their reactions to cessation-focused advertising based on their desire to quit, prior experience with quit attempts and, to a lesser degree, cigarette consumption. These are important considerations for campaign creators, designers and media planners.
Acknowledgments
The authors would like to thank Jennifer Duke for helpful comments on earlier drafts, Joshua Goetz for assistance with data analysis and Susan Murchie for editorial support.
References
Footnotes
Funding New York State Department of Health, Corning Tower, Room 710 Albany, NY 12237-0676, USA.
Competing interests None.
Ethics approval This study was conducted with the approval of the institutional review boards of RTI International and the New York State Department of Health.
Provenance and peer review Not commissioned; externally peer reviewed.