1 INTRODUCTION

Nationally, Australia has seen a decline in the prevalence of smoking over the last 30 years, commensurate with the sustained implementation of comprehensive tobacco control strategies, including tobacco price increases, the introduction of plain packaging, smoke-free places policy and legislation and the use of sustained and visible mass media campaigns.¹ In 2019, in New South Wales (NSW), Australia, 15.5% of adults were current smokers, compared to 19.4% of adults in 2008.² Notwithstanding these decreases, constant efforts are required to further increase cessation and to prevent a possible smoking resurgence³; this is of growing importance with recent research suggesting that the smoking rates have stabilized.²

It is known that mass media campaigns aired at regular intervals with sufficient levels of exposure in the population contribute to reducing smoking prevalence.^4-7 Specifically, campaigns that highlight the negative health consequences of smoking and evoke strong emotions, particularly those that use graphic imagery, are particularly effective.^{4, 8-10} With the link between tobacco control campaigns and decreasing smoking prevalence well established, there is a need to assess which emotional portrayals are most effective at increasing quitting-related activities and discouraging smoking to ensure the gains made by campaigns continue.

Evidence on the differential impact of positive and negative emotions evoking tobacco control campaigns must also be considered within the theoretical context of the role that emotions have in influencing perceptions, cognitions and behaviour.^11-13 For the last 30 years, theories have posited that people react to external or internal stimuli by appraising the harm or benefit of the stimuli before developing an emotional response which subsequently influences behaviour.^{11, 14} Many public health campaigns rely on the creation of an emotional response to a campaign message and that along with their perceived susceptibility, the severity of the risk and the threat of harm¹⁵ will result in change in the priming steps of behaviour change and the trialling of new behaviours. The ability of discrete emotions to trigger different behavioural responses is particularly relevant for campaigns, with research suggesting that fear tends to result in efforts to retreat from a threat and positively influence behaviour, sadness may provide a motivation¹⁰ to engage in immediate actions to ameliorate the sadness but may not result in long-term behaviour change^{13, 16} and hope providing a motivator for intentions and behaviours.¹⁷ Maximizing the ‘right’ emotional response is a crucial component in the anticipated success of a campaign and more recently, literature has suggested that consideration of how messages can stimulate a range of emotions (not just a dominant one) is required, noting that the emotion that the message evokes will evolve.¹²

The emotions evoked by tobacco control advertising can be classified as either positive (such as happiness and satisfaction) or negative (such as anger, fear, guilt, disgust or sadness).^{9, 10, 18-20} Evidence of differential effectiveness for campaigns that evoke negative or positive emotions is mixed for a range of outcomes, such as campaign recall, quit attempts, quitline calls and smoking prevalence.

In relation to campaigns that have elicited negative emotions; campaigns that evoke fear and sadness were seen as believable and thought provoking for recent quitters²¹; fear-evoking messages may also be important in informing people of the dangers of smoking²² and Richardson et al.¹⁹ reported that campaigns that elicited negative emotions had a greater impact on recall than positive emotion-evoking campaigns. Advertisements that elicited higher levels of guilt resulted in conversations that were more likely to contain talk of quitting by smokers.¹⁸ A recent study by Durkin et al.¹⁰ concluded that it might be more efficient to implement tobacco control advertisements that evoke fear rather than sadness and that advertisements that elicit a negative emotion may be most effective at motivating quitting amongst smokers from lower socio-economic backgrounds. On the other hand, research exploring the association between exposure to campaigns evoking different emotions and call volumes to smoking cessation quitlines found that campaigns evoking negative emotions did not influence call volumes.²³

The literature also provides details in relation to the use of both positive and negative emotion-evoking campaigns. The study by Durkin et al.¹⁰ also found that combining fear and hope-evoking campaigns leads to increased quit attempts among mid- high-socio-economic background smokers. Campaigns that evoke both positive and negative emotions have been shown to be effective in reducing smoking prevalence, with positive emotion campaigns playing a particular role in providing information on how to quit and a role in enhancing the self-efficacy of smokers.²⁴ Research from the American Tips From Former Smokers campaign is suggestive that the use of a wide variety of emotional appeals (including guilt, anger, fear, disgust, hope, warmth and humour), are effective at motivating smokers to make a quit attempt and nonsmokers to talk with friends and family about the harms associated with smoking.²⁰

In NSW, the Cancer Institute NSW (www.cancer.nsw.gov.au) is a state-level health agency that has developed and delivered comprehensive tobacco control campaigns on an annual basis since 2006. A range of campaigns highlight the negative health impacts associated with tobacco use and encourage smokers to quit smoking (Table 1), with the broadcasting of these campaigns associated with quitting outcomes at a population level.²⁵ This study builds on the evidence presented in the research undertaken by Durkin et al¹⁰ but with a different suite of campaign materials evoking a varied emotional response and in a different Australian jurisdiction. The current research explores campaigns (which included campaigns that predominantly evoked hope, sadness and combination of negative emotions) that the Cancer Institute NSW implemented during 2013–2018 and assessed associations between potential exposure to campaigns evoking specific emotions and quitting outcomes including: (i) quitting salience (ie, thinking about quitting); (ii) quit intentions; and (iii) quit attempts. In addition, this research assessed whether associations between quitting outcomes and potential campaign exposure differed according to demographic characteristics of smokers and recent quitters.

2 METHODS

2.2 Measures

2.2.1 Quitting-related outcomes

The CITTS collects information on smoking-related cognitions and behaviours, recognition and response to tobacco control campaigns and other policies, and quitting outcomes. Quitting salience (ie, thinking about quitting), was assessed by asking current smokers, ‘During the past two weeks, how often have you thought about quitting?’ With the response options being several times a day, once a day, once every few days, once a week and not at all. Responses were categorised into a binary indicator of whether or not respondents thought about quitting at least once per day in the 2 weeks before the survey.^{25, 26} Quitting intentions in the past month were determined by asking current smokers who indicated that they were ‘seriously considering quitting smoking in the next six months’, ‘Are you planning to quit smoking in the next 30 days?’. Those who were planning to quit within the next 30 days were classified as having quit intentions. In relation to quit attempts in the past month, smokers were asked ‘Have you ever tried to quit smoking before?’ and if they responded ‘Yes’, they were asked ‘How long ago did you last try to quit smoking?’ Responses were categorised into whether or not there was a quit attempt in the month before the survey. Recent quitters were asked ‘When did you finally stop smoking?’. Quit attempts were dichotomized as those who made a quit attempt in the month before the survey. If the recent quitter stopped smoking more than once month before the interview, they were excluded from these analyses.

2.2.2 Socio-demographic variables

Socio-demographic characteristics of survey participants included age, sex, individual-level socio-economic status (SES) and neighbourhood-level SES. A Heaviness of Smoking Index used a modified short form Fagerstrom questionnaire,²⁷ which uses a combination of ‘time to first cigarette after waking’ and ‘number of cigarettes smoked per day and divided into three categories (low, medium and high)’.²⁸ Change in cigarette price was assessed using the average price per cigarette per week during the study period.^{10, 29} Individual-level SES,²⁹ was assessed using a composite measure of household income and education.^{26, 30} High education refers to completing tertiary education (ie, University and TAFE); moderate education is completing the Higher School Certificate (ie, Year 12), and low education is not having completed the Higher School Certificate. High SES was defined as having a household income of more than AUD $80 000 (and any education level), or an income of AUD $40–80 000 and moderate-high education. Moderate SES was defined as either an income below AUD $40 000 and high education, or an income of AUD $40–80 000 and moderate education. Low SES was defined as either an income below AUD $40 000 and low or moderate education, or an income AUD $40–80 000 and low education. Neighbourhood-level SES was determined using postcode of residence and categorised using the Socio-Economic Index for Areas (SEIFA)^{25, 31} specifically the Index of Relative Socio-Economic Disadvantage, which ranks areas in Australia according to relative socio-economic disadvantage.

2.2.3 Campaign exposure

Potential exposure to campaigns was assessed in two ways (i) using the number of weeks (time) the campaign was run in the 3 months before the interview and (ii) dose: target audience rating points (TARPs) also in the 3 months before the interview. Potential exposure is a population-level proxy for direct individual-level exposure. TARPs represent the proportion of a specific target audience viewing a program at the time the advertisement is aired and as such are a product of the percentage of the target audience exposed to an advertisement, the average number of times it would be expected that a target audience member would be exposed to the advertisement.^{25, 32} TARPS are relevant for television campaigns; however, as not all of the campaigns in NSW between 2013 and 2018 were television campaigns (ie, Digital Quit smoking), and so we also examined potential exposure using time (ie, weeks that the campaign ran in the 3 months prior to the interview). Campaign exposure was calculated by summing the exposure (TARPS or weeks) during the 3 months before the interview for each campaign that was classified according to the dominant emotion evoked by that campaign (Table 1). Figure 1 provides a pictorial representation of the timing, and the emotional content of the tobacco control campaigns as they were ‘on air’ during the study period.

2.2.4 Classification of dominant emotions evoked by campaign

The dominant emotion evoked by each campaign was determined using a quantitative ratings study conducted by an external research agency on behalf of the Cancer Institute NSW. A sample (n = 152) of smokers and recent quitters were shown several advertisements (with each respondent being randomly allocated five advertisements to rate from a pool of 18 different advertisements) and asked to what extent each advertisement made them feel sad, angry, fearful, disgusted, guilty, hopeful and happy on a 5-point Likert scale ranging from ‘very’ to ‘not at all’. Consistent with previous research,¹⁰ the dominant emotion for each advertisement was coded if an emotion was evoked by at least 45% of the participants and there was no other emotion classification within 10%. When the emotion was evoked by at least 45% of participants, but other emotional classifications were within 10%, the advertisement was coded as ‘combined negative emotions’ which included a combination of fear, guilt and sadness (Table 1). No advertisements were classified as only evoking anger, fear, disgust, guilt, or happiness.

3 RESULTS

3.5 Interaction of hope and sadness and hope and negative emotions campaigns

There was a significant interaction between potential exposure (as measured by TARPS) to ‘hope’ and ‘sadness’ campaigns (F = 17.74, p = <.0001), such that thinking about quitting after exposure to ‘hope’ evoking campaigns decreased as exposure to ‘sadness’ campaigns increased. There was also a significant interaction between potential exposure (as assessed by TARPS) to ‘hope’ and ‘sadness’ campaigns and socio-economic status (F = 9.42, p = .009) (Figure 2) with thinking of quitting after exposure to ‘hope’ campaigns decreasing as exposure to ‘sadness’ campaigns increased among the high (F = 9.18, p < .001) and low (F = 16.86, p < .001) SES groups, but not the middle SES group.

Furthermore, stratified analyses indicated a significant interaction between ‘hope’ and ‘negative emotion’ TARPS in the moderate SES group (F = 5.10, p = .02), but not the high (F = 3.21, p = .07) or low (F = 0.53, p = .47) SES groups (Figure 2).

4 DISCUSSION

This study found a number of relationships between smokers' potential exposure to tobacco control mass media campaigns evoking different emotions and quitting outcomes. Specifically, smokers with greater potential exposure to campaigns that evoked ‘hope’ and campaigns that evoked ‘sadness’ were more likely to report salient quitting thoughts. Smokers potentially exposed to ‘hope’ campaigns were more likely to report quit intentions (in line with research undertaken by Nabi & Myrick,¹⁷) whereas potential exposure to ‘sadness’ or ‘negative emotion’ campaigns was not associated with quitting intentions. In terms of quit attempts, smokers who were potentially exposed to ‘sadness’ campaigns reported fewer quit attempts. Contrastingly, smokers potentially exposed to ‘negative emotion’ campaigns were most likely to report recent quit attempts, in line with existing research.^{10, 19, 24}

‘Negative emotion’ campaigns evoked three negative emotions: fear, guilt and sadness. Our findings broadly support that these emotions are associated with quit attempts. This finding is consistent with the study undertaken by Durkin et al¹⁰ who found that exposure to advertisements which evoked multiple negative emotions (including fear, guilt and sadness) increased the likelihood of past-month quit attempts (particularly for lower SES areas). However, our study did not find any differential impacts for exposure to combined negative emotions on quitting attempts between SES groups. Many of these negative emotion-framed campaigns utilised graphic advertisements depicting health consequences of smoking; previous research³³ has demonstrated the importance and influence of this type of anti-tobacco campaign on quit attempts.

In addition to potential exposure to ‘sadness’ campaigns having a negative association with quit attempts, our results suggest an interaction between ‘hope’ and ‘sadness’ evoking campaigns, and that exposure to a sadness campaign may offset the positive impact of a ‘hope’ campaign on quitting salience. This finding supports the hypothesis posed by Durkin et al¹⁰ that exposure to predominantly sadness-evoking advertisements might reduce the likelihood of quitting. It seems that there is a risk that sadness campaigns may lead to feelings of hopelessness and helplessness in smokers, and that rather than bringing about quitting related action they may result in smokers not taking any action, and avoiding seeking help to quit^{11, 16, 34}; this explanation is also in keeping with theories regarding the use emotional focused coping which tends to be used when there is an appraisal that nothing can be done to modify or eliminate the problem.^{10, 13, 16, 35} Campaigns that were classified as evoking ‘sadness’ tended to highlight the regret at missing out of time or experiences with loved ones which may have a stronger impact on feelings of helplessness.³⁶

In this study, potential exposure to ‘hope’ campaigns had a particular resonance with sub-groups of smokers, potential exposure to such campaigns were associated with quitting salience for heavy smokers and with quit intentions and quit attempts for older smokers. Other research suggests that positive emotion campaigns may be important to enhance self-efficacy when used in combination with negative evoking campaigns.²⁴ Our study found that thinking about quitting after potential exposure to ‘hope’ campaigns decreased as potential exposure to ‘sadness’ increased, and this effect was moderated by SES. While this finding requires greater exploration to understand how best to combine positive and negative evoking campaigns, it may be that campaigns that unduly evoke ‘sadness’ could weaken the impact of campaigns that evoke positive emotions although this may be related to the particular type of sadness evoking campaigns.

The results presented here confirm that the emotions evoked by a campaign are important in designing and planning tobacco-reduction messages and campaigns.³⁷ Evoking a positive emotion such as hope is useful but likely not sufficient to bring about actual quit attempts; whereas a negative emotion-provoking campaign does seem to increase the likelihood of quit attempts but does not seem to influence quitting cognitions or quitting plans. This suggests that population-wide campaigns should target negative emotional responses, particularly fear and guilt,²⁴ to increase quit attempts. There may be a role for the more targeted use of hopeful campaigns aimed at older and heavier smokers. Suggesting that older and heavier smokers may have a number of failed quit attempts and that they require hopeful messages to increase their self-esteem. Furthermore, hope campaigns might overcome cognitive dissonance regarding quitting and quitting intentions, but by themselves do not provide the best opportunity to influence quit attempts, although there is merit in the use of such messages targeting older and heavier smokers. There may also be a role for hope-evoking campaigns that target smokers to influence knowledge, attitudes and beliefs as priming steps of behaviour change and quit attempts. Further research will be needed to determine the ideal combination of these campaigns. Notwithstanding that the advertisements included in our study only evoked hope, sadness and combined negative emotions, as tobacco control efforts target further declines in smoking, it may also be opportune to assess campaigns that induce a broader range of feelings (for example hypocrisy³⁸) on smoking-related behaviours.

5 LIMITATIONS AND STRENGTHS

These results should be interpreted within NSW context of several styles and types of tobacco control advertising since 1983 and it is not possible to account for this exposure in this analysis. Nor does the analysis account for the creative execution or type of media placement, including the more recent integration of digital media.³⁹ Our study does not account for other unobserved noncampaign factors (social and individual factors, illness etc.) that might influence decisions to quit smoking, although our analysis did adjust for tobacco price and heaviness of smoking. These surveys are based on self-report, but since they are estimating psychological intermediates, they are considered acceptable, but still may be subject to social desirability bias. Further limitations of this paper include: the study in the main focuses on a sub-sample of CITTS respondents; it uses potential exposure to the campaign materials rather than actual exposure and must be acknowledged as a limitation of the study design, and a separate study was used to assess the dominant emotion of the campaigns rather than a sub-sample of the main study.

A strength of our study is that it included multiple quit-related behaviours and antecedents, including thinking about quitting, quit intentions, and quit attempts, allowing an examination of the relative effects of different types of campaigns on a range of tobacco campaign endpoints. The study is sample weighted to represent the NSW population, thus increasing the generalisability of results.

6 CONCLUSIONS

Echoing a sentiment expressed by others, further research¹⁰ is needed internationally on similar campaigns to examine if findings differ or are consistent with these Australian findings. The emotion that a campaign evokes in smokers should be an important consideration in the design of future tobacco control campaigns and as suggested by our study utilising a mix of campaigns that evoke both positive and negative emotions depending on the purpose and target of the campaign is seen as critical in maintaining campaign effectiveness.

AUTHOR CONTRIBUTIONS

All authors contributed to the idea of the study and to the interpretation of the results. BOH led the implementation of the study and wrote the first draft of the article, KO led the analysis and all authors reviewing and approving the final version.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

ETHICS STATEMENT

Ethics approval for this study was granted by the University of Sydney Human Research Ethics Committee (project number: 2017/497).