Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Oct 10.
Published in final edited form as: Addict Behav. 2017 Apr 4;73:22–29. doi: 10.1016/j.addbeh.2017.04.002

The role of negative affect and message credibility on perceived effectiveness of smokeless tobacco health warning labels in Navi Mumbai, India and Dhaka, Bangladesh: A moderated-mediation analysis

Seema Mutti-Packer a, Jessica L Reid b, James F Thrasher c,d, Daniel Romer e, Geoffrey T Fong a,f,g, Prakash C Gupta h, Mangesh S Pednekar h, Nigar Nargis i, David Hammond a
PMCID: PMC6786900  NIHMSID: NIHMS985686  PMID: 28441573

Abstract

Objective

There is strong evidence showing that pictorial health warnings are more effective than text-only warnings. However, much of this evidence comes from high-income countries and is limited to cigarette packaging. Moreover, few studies have identified mechanisms that might explain the impact of warnings.

Methods

The current study examined the potential mediating role of negative affect and the moderating influence of message credibility on perceived effectiveness of smokeless tobacco warnings in two low- and middle-income countries (LMICs). Field interviews were conducted in India and Bangladesh, with adult (19+ years) smokeless tobacco users (n=1,053), and youth (16–18 years) users (n=304) and non-users (n=687). Respondents were randomly assigned to view warnings in one of four conditions: (1) Text-only, (2) Pictorial with symbolic imagery, (3) Pictorial with graphic images of health effects, or (4) Pictorial with personalized graphic images plus a personal testimonial.

Results

The findings provide support for the mediating influence of negative affect on perceived effectiveness, for adult and youth smokeless tobacco users who viewed pictorial warnings (vs. text-only), and graphic health warnings (vs. personal testimonials). Among adults, message credibility moderated the indirect effect; the association was stronger when credibility was high and weaker when it was low. Among youth users and non-users, message credibility did not moderate the indirect effect.

Conclusions

Consistent with research from high-income countries, these findings highlight the importance of selecting imagery that will elicit negative emotional reactions and be perceived as credible. Differential effects among adults and youth highlight the importance of pre-testing images.

Keywords: smokeless tobacco, health warnings, low and middle income countries, negative affect, message credibility

1. INTRODUCTION

Globally, smokeless tobacco use is disproportionately concentrated in India and Bangladesh. These two countries account for about 80% of the approximately 300 million smokeless tobacco users worldwide (National Cancer Institute, 2014). Despite considerable evidence linking smokeless tobacco use with oral cancer and other adverse health outcomes (International Agency for Research on Cancer, 2007) knowledge of the health risks of smokeless tobacco use remains low in the Indian subcontinent, as well as in migrant South Asian communities worldwide (Gupta & Ray, 2003; Kakde & Jones, 2012; Khawaja et al., 2006; Rahman et al., 2012; Messina et al., 2013). Indeed, India has one of the highest incidences of oral cancer in the world (Khan, 2012).

Health warnings on product packaging are one example of a cost-effective, population-wide strategy to inform populations about the health risks associated with tobacco use (Hammond, 2011). The World Health Organization’s Framework Convention on Tobacco Control (WHO FCTC) established international standards for packaging and health warnings: FCTC Article 11 Guidelines call for pictorial warnings covering at least 50% or more of the pack (World Health Organization, 2008). India and Bangladesh, both signatory countries to the FCTC, differ with respect to their tobacco control environments. Despite numerous delays in implementation (Arora, Tewari, Nazar, Gupta & Shrivastav, 2012; Oswal, Raute, Pednekar, & Gupta, 2011), in 2009, India became the first country in the world to require pictorial health warnings (a symbolic image of a scorpion) on smokeless tobacco packages. In 2011, the symbolic image of a scorpion was replaced with four different graphic images, but recent longitudinal evidence indicated that this change from a symbolic warning in 2009, to graphic warnings in 2011, did not result in significant increases in effectiveness (Gravely et al., 2016). However, India has committed to increasing warning size from the current 40% of the front surface to 85% of the principal display area (i.e., front and back). As of March 2016, Bangladesh has implemented pictorial health warnings covering 50% on smokeless tobacco packages1.

In terms of message content, there is strong evidence supporting the superiority of pictorial health warnings over text-only warnings in promoting smoking cessation, and increasing health knowledge and perceptions of risk (Hammond, 2011; Noar et al., 2015). Pictorial warnings often contain graphic, fear-arousing images that elicit negative emotion. Empirical research suggests that strong, fear-arousing messages are most likely to alter beliefs about health risks, as well as appeal and general acceptability of tobacco products (Emery, Romer, Sheerin, Jamieson, & Peters, 2014; Evans et al., 2015; Netemeyer, Burton, Andrews, & Kees, 2016; Hammond, 2011; Noar et al., 2015). Alongside gruesome images of disease caused by tobacco use, graphic health warnings may also include narratives, such as personal “testimonials” from tobacco users. However, findings are mixed with respect to the efficacy of testimonials on health warnings (Hammond et al., 2012; Thrasher et al., 2012).

With few exceptions, much of this research has been conducted in high-income countries and is almost entirely based on cigarette package warnings. To our knowledge, only three studies (one from the US, one from Canada, and one based on the current dataset), have experimentally tested attributes of health warnings for smokeless tobacco (Adkison, Bansal-Travers, Smith, O’Connor, & Hyland, 2014; Callery, Hammond, O’Connor, & Fong, 2011; Mutti et al., 2015). The current study is a follow-up analysis to an experimental study conducted by Mutti et al. (2015), which examined the perceived effectiveness of health warning labels with different message content for smokeless tobacco packages in India and Bangladesh. The previous analysis found that text-only messages were rated as less effective than all of the styles of warnings tested, including warnings with symbolic imagery, graphic health effects, and personal testimonials. Further, among the pictorial style warnings, graphic health warnings were rated as more effective than symbolic and personal testimonial warnings. Overall, the findings from this and other studies on smokeless tobacco health warnings support previous findings based on health warnings for cigarette packaging; pictorial warnings are more effective than text-only warnings, and warnings including graphic images are most effective overall.

Despite these promising findings, to date, no studies have explicitly examined the role of negative emotion or other factors that might determine the effectiveness of smokeless tobacco warnings in developing countries. Thus, while the central question of whether graphic images of health effects are the best approach for smokeless tobacco warnings is starting to be addressed, the secondary question of whether this effect is mediated or moderated by other underlying factors has yet to be examined in the context of LMICs. Overall, the evidence from studies based on cigarette health warnings indicates that cognitive and affective mediators underlie the effectiveness of health warnings. These studies have consistently found that compared to text-only warnings, warnings with graphic images elicited greater fear arousal, which in turn increased intentions to quit (Kees, Burton, Andrews, & Kozup, 2010), perceived risk (Emery et al., 2014), as well as perceived and actual effectiveness (Byrne, Katz, Mathios, & Niederdeppe, 2015).

According to the Extended Parallel Process Model (Witte, 1992), it is possible that a high level of fear arousal could result in message rejection. In addition, dual-process theories of attitude change such as the Elaboration Likelihood Model (Petty & Cacioppo, 1986) underscore the importance of engaging not only affective pathways, but also cognitive pathways by ensuring that warnings are credible and believable (Strahan et al., 2002). Thus, it is possible that a highly credible message could enhance warning label effectiveness, as well as diffuse heightened emotional responses that might otherwise lead to message rejection. To our knowledge, only one experimental study examined the role of message credibility as a moderator of warning label impact (Emery et al., 2014). Based on a sample of smokers, Emery and colleagues found that message credibility interacted with affect in predicting a positive attitude towards quitting. Additional studies have examined message credibility as an outcome (Thrasher et al., 2012) and as a mediator (Evans et al., 2015). Overall, negative affect and message credibility appear important in understanding how cigarette warnings work, and are likely to apply to smokeless tobacco package health warnings as well.

The primary aim of the current study was to extend the analysis conducted by Mutti and colleagues (2015) by examining potential affective and cognitive factors that may influence the warnings’ effectiveness. Based on the fear appeal literature, as well as previous findings, it was hypothesized that negative affect would mediate the association between health warning type and ratings of perceived effectiveness; warnings which elicit higher levels of negative affect would elicit higher perceived effectiveness ratings. Therefore, based on the original study (Mutti et al., 2015), it was expected that all of the pictorial styles (symbolic, graphic, and personal testimonial) would elicit greater levels of negative affect compared to text-only warnings. Furthermore, it was expected that graphic health warnings would elicit greater negative affect compared to warnings with personal testimonials. (H1). It was also hypothesized that the association between negative affect and perceived effectiveness would vary as a function of message credibility; the association would be stronger when message credibility was high, and weaker when it was low (H2).

2. METHODS

An experimental study was conducted in India (n=1,002) and Bangladesh (n=1,081), with adult (19+ years) smokeless tobacco users, and youth (16 to 18 years) users and non-users. Adults and youth represent critical sub-groups in which to evaluate perceptions of health warnings, given that warnings may have differential effects with respect to promoting cessation among current smokeless tobacco users and discouraging uptake of smokeless tobacco use among youth. Ethical clearance was received from the University of Waterloo Office of Research Ethics, the Healis-Sekhsaria Institute for Public Health and the Bangladesh Medical Research Council. Equal proportions of male and female respondents were recruited; no quota limits were placed on users and non-users. Respondents were recruited using the intercept technique (Sudman, 1980), whereby a physical landmark was selected, and every other person to pass it was approached in Navi Mumbai, India. In Dhaka, Bangladesh, every third person to pass the landmark was approached in busy locations, which was increased to every person in locations with less pedestrian traffic. After verbal consent was given, trained interviewers conducted face-to-face interviews using tablets. Interviewers read questions aloud to respondents and entered their responses on tablets.

2.1. Measures

2.1.1. Multi-categorical predictor variable: Message theme

A between-subjects design was used, randomly assigning respondents to view smokeless tobacco health warnings in one of four experimental conditions: (1) Text-only (2) Pictorial with symbolic imagery representing danger or caution, (3) Pictorial with a graphic health effect depicting a diseased body part/organ, or (4) Pictorial with a personalized graphic image of a “real” person, with an accompanying personal testimonial. Please see online Supplementary Figure for all health warnings tested in this study. Each respondent viewed five warnings within that condition, one for each of the following health effects: oral cancer, mouth disease, heart disease, addiction, and death. All warnings included the text “TOBACCO KILLS” at the bottom2. Condition 1 warnings consisted of five text-only warnings corresponding to each of the five health effects (i.e., “Tobacco causes oral cancer”, “Tobacco causes mouth disease”, etc.). The same text, corresponding to the five health effects, was used in all conditions. In Condition 1 (text-only), this text occupied the main panel of the warning, while Conditions 2, 3, and 4 featured images, and Condition 4 also included personal testimonial text with name and age. Warnings within each set were shown in random order on a tablet screen, and rated one at a time, while each warning appeared onscreen. In India, approximately 125 respondents were allocated to each of the four experimental conditions (502 adults, and 500 youth). In Bangladesh, approximately 130 respondents were allocated to each condition (569 adults, and 512 youth). For a full description of the methodology, see Mutti et al., 2015.

2.1.2. Mediator and moderator: Negative affect and message credibility

Using a numeric scale, from 1 to 10, with anchors 1=“not at all”, 5=“in the middle”, and 10=“extremely”, warnings were rated on the following measures: “Please tell me whether this warning message…” “is believable”; “is surprising”; “is frightening”; “is disgusting”; and, “is unpleasant”. The item “is believable” was used as the measure of credibility. The four measures conceptualized as ‘negative affect’ (fright, disgust, surprise, and unpleasant) were highly correlated across all experimental conditions (Cronbach’s α ranged from 0.97 to 0.98). No differences were observed when each of these measures were tested separately. Thus, the Negative Affect Scale was created, whereby these four measures were summed across the five warnings within each experimental condition and then divided by twenty, yielding a mean score ranging from 1 to 10 for each condition.

2.1.3. Outcome variable: Perceived effectiveness

Using the same question root and scale, four items measured perceived effectiveness, including whether the warning would: 1) “make people more concerned about the health risks of using smokeless tobacco”; 2) “help prevent young people from starting to use smokeless tobacco”; and 3) “make smokeless tobacco users want to quit”; as well as, 4) “Overall, on a scale of 1 to 10, how effective is this health warning?” These four measures were highly correlated across experimental conditions (Cronbach’s α=0.95 to 0.97); thus, only the measure of “overall effectiveness” was used in the analysis. The “overall effectiveness” measure was summed across the five warnings within each experimental condition and then divided by five (number of warnings in each condition), yielding a mean score between 1 and 10 for each condition.

2.1.4. Covariates

The following covariates were entered into the model examining the adult sample: country, age, sex, education, income, smokeless tobacco use (daily users and nondaily users), mixed use (i.e., use of both smokeless and smoked tobacco), and quit intentions (any intention to quit vs. none). Covariates in the model examining youth smokeless tobacco users included country, age, sex, and smokeless tobacco use (daily users and nondaily users), mixed use, and quit intentions. For youth non-users covariates included country, age, sex, and susceptibility to smokeless tobacco use (non susceptible vs. susceptible) (Pierce et al., 1996). Covariates were added to all paths (a,b, and c’ in Figure 1) in the model. In interviews, all covariates were assessed prior to viewing the health warnings. For a full description of the measures used, refer to Mutti et al., 2015.

Figure 1.

Figure 1.

Open in a new tab

Conceptual moderated-mediation model with a multi-categorical predictor variable*

2.2. Analytic approach

Chi-square tests (for categorical variables), one-way ANOVA and t-tests (for continuous variables) were conducted to examine differences between the Indian and Bangladeshi samples, using SPSS version 24.0. Mediation and moderated-mediation analyses were conducted using the SPSS PROCESS macro, developed by Hayes (2013). The PROCESS macro is designed to test conditional process models using ordinary least squares regression-based path analysis to test for direct and indirect effects. The predictor variable was multi-categorical, with four levels: (1) text-only, (2) symbolic, (3) graphic health effect, and (4) personal testimonial3. Respondents with missing data on any of these covariates or outcome measures (n=39) were excluded from analyses on a list-wise basis.

To examine whether negative affect mediated the effect of health warning content on perceived effectiveness ratings, mediation analyses were conducted (Model 4, PROCESS macro). The indirect effect (mediation) was estimated with bias-corrected (BC) 95% confidence intervals (CI) of 5,000 bootstrapped samples (Hayes, 2013; Preacher & Hayes, 2004). For statistical inference of indirect effects (mediation), confidence intervals that do not include zero indicate significance. To examine whether message credibility moderated the indirect effect, the interaction term of negative affect by message credibility was included in the model (Model 14, PROCESS macro). For the moderator of message credibility, significant interactions were probed further by using the ‘pick-a-point approach’, meaning that the conditional indirect effect of the moderator was examined at one standard deviation (SD) above and below the mean; levels corresponded to ‘low’, ‘moderate’, and ‘high’ levels of message credibility (Hayes, 2013). Figure 1 presents the conceptual moderated-mediation model with a multi-categorical predictor variable.

3. RESULTS

3.1. Sample characteristics

Table 1 presents the characteristics of the adult and youth samples, by country, noting significant differences between countries.

Table 1.

Characteristics of the adult and youth samples in Navi Mumbai, India and Dhaka, Bangladesh

ADULTS (n=1,071) YOUTH NON-USERS (n=704) YOUTH USERS (n=308)
Indiay (n=502) Bangladesh (n=569) Test statistic† (p-value) India (n=326) Bangladesh (n=378) Test statistic† (p-value) India (n=174) Bangladesh (n=134) Test statistic† (p-value)
Age (mean; SD) 36.0 (9.2) 38.6 (12.5) t=3.8 (p<0.001) 17.4 (0.7) 17.1 (0.8) t=−5.9 (p<0.001) 17.3 (0.8) 17.6 (0.6) t=−4.0 (p<0.001)
Sex (n; %)
 Female 250 (49.8) 261 (45.9) X2=1.6 (p=0.22) 185 (56.7) 222 (58.7) X2=0.28 (p=0.65) 65 (37.4) 32 (23.9) X2=6.4 (p=0.01)
 Male 252 (50.2) 308 (54.1) 141 (43.4) 156 (41.3) 109 (62.6) 102 (76.1)
Smokeless tobacco use (n; %)
 Daily user 470 (93.6) 537 (94.4) X2=0.3 (p=0.61) -- -- 145 (83.3) 74 (55.2) X2=29.1 (p<0.001)
 Non-daily user 32 (6.4) 32 (5.6) -- -- 29 (16.7) 60 (44.8)
 Non-user susceptible -- -- 106 (32.5) 79 (20.9) X2=12.2 (p=0.001) -- --
 Non-user non-susceptible -- -- 220 (67.5) 299 (79.1) -- --
Mixed use (n; %)* 85 (16.9) 141 (24.8) X2=9.9 (p=0.002) -- -- 32 (18.4) 29 (21.6) X2=0.50 (p=0.56)
Quit intentions (n; %)
 Plans to quit 350 (69.7) 284 (50.1) X2=42.5 (p<0.001) -- -- 142 (81.6) 66 (49.6) X2=35.3 (p<0.001)
 Not plans to quit 152 (30.3) 283 (49.9) -- -- 32 (18.4) 67 (50.4)
Income (n; %)
 Low 193 (38.5) 412 (72.8) X2=131.0 (p<0.001) -- -- -- --
 Moderate 175 (34.9) 101 (18.0) -- -- -- --
 High 53 (10.4) 17 (3.0) -- -- -- --
 Not stated 81 (16.2) 35 (6.2) -- -- -- --
Education (n; %) -- -- -- --
 Low 19 (3.8) 179 (31.5) X2=247.2 (p<0.001) -- -- -- --
 Moderate 223 (44.4) 316 (55.6) -- -- -- --
 High 260 (51.8) 73 (12.9) -- -- -- --
Open in a new tab
†

Test statistic denotes between-country differences (India vs. Bangladesh), within adult or youth user/non-user sample. Significant differences bolded.

*

Represents respondents who reported using both smoked and smokeless products.

Missing: Quit Intentions (Bangladeshi Adults, n=2; Bangladeshi youth users, n=1); Income (Indian adults, n=1; Bangladeshi adults, n=4); Education (Bangladeshi adults, n=1). For a full description of the measures used, please refer to Mutti et al., 2015.

3.2. The mediating influence of negative affect

As shown in Table 2, for adults and youth, viewing any style of the pictorial style warnings (symbolic, graphic, or testimonial) elicited greater levels of negative affect compared to text-only warnings (path a in Figure 1). In addition, for adult and youth smokeless tobacco users, viewing graphic warnings elicited greater levels of negative affect compared to viewing testimonial warnings. In contrast, among youth non-users, viewing symbolic warnings did not elicit significantly higher levels of negative affect compared to text-only warnings, nor did viewing graphic warnings compared to personal testimonials. However, across all subsamples, a significant positive association was found for path b, indicating that greater levels of negative affect elicited greater levels of perceived effectiveness.

Table 2.

Regression coefficients for mediation analyses examining the effect of message theme (X) on perceived effectiveness (Y) via negative affect (M), in India and Bangladesh
ADULTS (n=1,053) YOUTH USERS (n=304) YOUTH NON-USERS (n=687)
Path a
X→ Negative affect (M) X→ Negative affect (M) X→ Negative affect (M)
MESSAGE THEME (X) Coeff. (SE) p value Coeff. (SE) p value Coeff. (SE) p value
Symbolic (v. text) 0.48 (0.17) 0.006 0.55 (0.25) 0.03 0.08 (0.19) 0.68
Graphic (v. text) 2.14 (0.18) <0.001 2.55 (0.24) <0.001 2.48 (0.19) <0.001
Testimonial (v. text) 1.28 (0.18) <0.001 1.82 (0.24) <0.001 2.10 (0.19) <0.001
Graphic (v. testimonial) 0.86 (0.18) <0.001 0.72 (0.24) 0.002 0.37 (0.18) 0.05
Path b
M→ Perceived Effectiveness (Y) M→ Perceived Effectiveness (Y) M→ Perceived Effectiveness (Y)
Coeff. (SE) p value Coeff. (SE) p value Coeff. (SE) p value
NEGATIVE AFFECT (M) 0.51 (0.02) <0.001 0.58 (0.03) <0.001 0.51 (0.02) <0.001
Path c Path c’ Path c Path c’ Path c Path c’
X→ Perceived Effectiveness (Y) X→ Perceived Effectiveness (Y) X→Perceived Effectiveness (Y) X→Perceived Effectiveness (Y) X→Perceived Effectiveness (Y) X→ Perceived Effectiveness (Y)
MESSAGE THEME (X) Coeff. (SE) p value Coeff. (SE) p value Coeff. (SE) p value Coeff. (SE) p value Coeff. (SE) p value Coeff. (SE) p value
Symbolic (v. text) 0.35 (0.13) 0.007 0.10 (0.09) 0.28 0.43 (0.20) 0.032 0.11 (0.14) 0.43 0.21 (0.15) 0.18 0.17 (0.12) 0.16
Graphic (v. text) 2.24 (0.13) <0.001 1.14 (0.10) <0.001 2.53 (0.19) <0.001 1.05 (0.15) <0.001 2.63 (0.16) <0.001 1.36 (0.14) <0.001
Testimonial (v. text) 1.70 (0.13) <0.001 1.05 (0.10) <0.001 1.88 (0.19) <0.001 0.83 (0.14) <0.001 2.25 (0.15) <0.001 1.17 (0.13) <0.001
Graphic (v. testimonial) 0.53 (0.13) <0.001 0.09 (0.09) 0.34 0.64 (0.19) 0.007 0.23 (0.13) 0.09 0.37(0.15) 0.15 0.19 (0.12) 0.12
INDIRECT EFFECT OF X ON Y (Test of mediation)
Symbolic (v. text) 0.25 (Bias-corrected 95% CI: 0.06, 0.44) 0.32 (Bias-corrected 95% CI: 0.03, 0.61) 0.04 (Bias-corrected 95% CI: −0.14, 0.21)
Graphic (v. text) 1.10 (Bias-corrected 95% CI: 0.90, 1.33) 1.47 (Bias-corrected 95% CI: 1.14, 1.86) 1.27 (Bias-corrected 95% CI: 1.01, 1.55)
Testimonial (v. text) 0.66 (Bias-corrected 95% CI: 0.48, 0.86) 1.05 (Bias-corrected 95% CI: 0.74, 1.40) 1.07 (Bias-corrected 95% CI: 0.84, 1.35)
Graphic (v. testimonial) 0.44 (Bias-corrected 95% CI: 0.28, 0.62) 0.42 (Bias-corrected 95% CI: 0.16, 0.71) 0.19 (Bias-corrected 95% CI: −0.01, 0.40)
Open in a new tab

Notes. Unstandardized regression coefficients presented in table. X=independent variable, M=mediating variable; Y=dependent variable. Path a= X →M; Path b= M→Y; Path c (total effect)= X→Y; Path c’ (direct effect)= X→Y, controlling for M. Indirect effect of X on Y (mediation) is the product of Path a and Path b.

Results shown are from the final model including all covariates. R2 change between the baseline model (no covariates in model) and the final conditional model (including covariates): Adults R2 baseline model=0.60, R2 final model=0.67, R2 change=0.07; Youth users R2 baseline model=0.74, R2 final model=0.75, R2 change=0.01; Youth non-users: R2 baseline model=0.61, R2 final model=0.64, R2 change=0.03.

Bias-corrected 95% confidence intervals (CI) that do not contain zero indicate significance of the indirect effect (mediation).

With respect to the inferential test of the indirect effect (the product of Path a and Path b), the findings suggest that negative affect mediated the effect of viewing health warnings and ratings of perceived effectiveness across all message theme contrasts, for adults and youth smokeless tobacco users. Among youth non-users, a mediating effect was not present for those who viewed symbolic warnings, compared to text-only, nor for those who viewed graphic warnings compared to testimonials.

Full mediation occurred among adults, and youth smokeless tobacco users. Specifically, the effect of viewing graphic health warnings (vs. personal testimonials) on ratings of perceived effectiveness became non-significant when controlling for negative affect, indicating that the effect was explained entirely by the mediating variable (Adults Path c: 0.53, p<0.001 to Path c’: 0.09, p=0.34; Youth users Path c: 0.64, p=0.007 to Path c’: 0.23, p=0.09). In addition, the effect of viewing symbolic warnings (vs. text-only) on ratings of perceived effectiveness became non-significant when controlling for negative affect, for both adults and youth users (Adults Path c: 0.35, p=0.007 to Path c’: 0.10, p=0.28; Youth users Path c: 0.43, p=0.03 to Path c’: 0.11, p=0.43). For all other significant indirect effects, Path c to Path c’ was reduced but remained significant, indicating partial mediation.

3.3. The moderating influence of message credibility

To examine the potential moderating influence of message credibility, the interaction term of negative affect by message credibility was added to the model. Message credibility was found to moderate the indirect effect in the adult sample, but not among the younger sample.

Among adults, this interaction term was significant (β=0.045, SE=0.007, p<0.001), and confirmed by a formal test using the index of moderated mediation (0.02 (Bias-corrected 95% CI: 0.01, 0.04)), indicating that the mediating influence of negative affect on perceived effectiveness ratings, varied as a function of message credibility. The interaction term was not significant among youth (Users: β=−0.003, SE=0.014, p=0.85; Non-users: β=−0.001, SE=0.008, p=0.89), and was confirmed by the index of moderated mediation (Users: −0.001 (Bias-corrected 95% CI: −0.02, 0.01); Non-users: −0.001 (Bias-corrected 95% CI: −0.004, 0.002)).

The significant interaction from the adult sample was probed to assess the conditional indirect effect, at three levels of the moderator (message credibility): one SD below the mean, the mean, and one SD above the mean—corresponding to ‘low’, ‘moderate’, and ‘high’ levels of credibility. Table 3 presents the results from this analysis. Overall, the results in Table 3 indicate that the indirect effect was stronger when message credibility was high and weaker when it was lower, among adults.

Table 3.

Conditional indirect effects of message theme (X) on perceived effectiveness ratings (Y) via negative affect (M), at different values of the moderator among adults in India and Bangladesh (n=1,053)

Credibility (V) Coefficient (bootstrapped SE) Bias-corrected 95% bootstrapped CI
Symbolic (v. text)
Low Credibility 0.14 (0.06) 0.04 – 0.26
Moderate Credibility 0.18 (0.07) 0.05 – 0.33
High Credibility 0.23 (0.09) 0.07 – 0.41
Graphic (v. text)
Low Credibility 0.62 (0.10) 0.43 – 0.84
Moderate Credibility 0.82 (0.10) 0.64 – 1.04
High Credibility 1.02 (0.11) 0.82 – 1.26
Testimonial (v. text)
Low Credibility 0.37 (0.08) 0.24 – 0.54
Moderate Credibility 0.49 (0.08) 0.35 – 0.66
High Credibility 0.61 (0.09) 0.45 – 0.80
Graphic (v. testimonial)
Low Credibility 0.25 (0.06) 0.14 – 0.28
Moderate Credibility 0.33 (0.07) 0.20 – 0.48
High Credibility 0.41 (0.09) 0.25 – 0.59
Open in a new tab

The three levels of the moderator correspond to one SD below the mean, the mean, and one SD above the mean. Thus, Low credibility=4.37, Moderate credibility=6.48, High credibility=8.59.

Bias-corrected 95% confidence intervals (CI) that do not contain zero indicate significance of the conditional indirect effect (moderated-mediation).

4. DISCUSSION

Overall, the findings support the broader literature suggesting that negative emotions such as fear underlie the effectiveness of warnings (Wang et al., 2015; Hammond Fong, McDonald, Brown, & Cameron, 2004; Borland et al., 2009; Peters et al., 2007; Witte and Allen, 2000; BRC Marketing and Social Research, 2004; Elliot & Shanahan Research, 2003; Environics, 1999, 2000; Kees et al., 2010). The findings on the desirable effects of negative emotional responses to warnings are also in line with recent experimental field studies (Evans et al 2015) and observational research (Cho et al 2016), including research based on nationally representative longitudinal data from four high-income countries (Yong et al., 2014). Yong and colleagues (2014) used mediational analyses to identify factors that explained the impact of warnings on key indicators of cessation among adult smokers. They found that large, graphic warnings work indirectly through several underlying factors: they encouraged smokers to think about the risks of smoking, which elicited emotional responses such as worry about health risks, which in turn increased intentions to quit and encouraged subsequent quitting behavior. In addition, contrary to the idea that extreme fear arousal results in message rejection (Kessels, Ruiter, Wouters, & Jansma, 2014; Kessels, Ruiter, & Jansma, 2010), avoidance behavior was associated with increased frequency of thinking about health risks (Yong et al 2014) and subsequent cessation attempts (Thrasher et al 2016). Together, these findings support the use of graphic, fear-arousing tobacco health warnings.

Mediation analyses revealed that smokeless tobacco users, both adults and youth, responded similarly to warnings. The finding related to youth is of particular importance given the limited research focused on this age cohort, as well as previous mixed findings with respect to the effectiveness of pictorial warnings among youth (Pepper et al., 2013; Noar et al., 2015). However, for non-using youth, a mediating effect was not found among those who viewed symbolic warnings (vs. text), nor for those who viewed graphic warnings (vs. personal testimonials), indicating potential differential effects between youth smokeless tobacco users and non-users. We can only speculate as to the mechanism driving this effect, but one explanation might be that for non-using youth, there may be a lack of immediacy of the risk compared to youth users.

Although the results provide support for the potential mediating role of negative affect for both adults and youth, support for the second hypothesis regarding the moderating effect of message credibility on the association between negative affect and perceived effectiveness was found only among adults. Specifically, the strength of the association between negative affect and perceived effectiveness varied as a function of message credibility across all message theme contrasts, even when comparing graphic health warnings and personal testimonials. This finding is consistent with research based on a sample of smokers who were assigned to view warnings in one of three conditions: text-only, graphic image with basic text, and a graphic image with elaborated text. In their study, Emery and colleagues (2014) found that message credibility interacted with affect in predicting a positive attitude towards quitting.

One possible explanation for the finding that message credibility did not moderate the indirect effect for youth could be that graphic health warnings and warnings with personal testimonials featured older individuals, thus rendering the health effects of smokeless tobacco as too remote for youth, and not as believable. Furthermore, it is possible that these younger respondents may have not been able to identity with the depicted individual (Kreuter et al., 2007; Larkey & Hecht, 2010), nor with the experience conveyed (i.e, the death of a spouse in the personal testimonial condition—see Supplementary Figure). Future research might consider examining this potential issue of personal relevance, in addition to message credibility. Overall, this set of findings related to the differential effects among adults and youth, and users and non-users, highlights the need to pre-test images with the target audience.

Overall, the findings from the current study demonstrate the mechanism underlying the effect established in the original study by Mutti and colleagues (2015); that is, fear and message credibility underlie the perceived effectiveness of graphic health warnings when compared to all other warning styles, including personal testimonials. However, it is important to note that the true efficacy of personal testimonials may have been underestimated. Personal testimonials may not have the same hard-hitting impact as warnings showing graphic health effects. The images used for personal testimonial warnings in this study depicted a ‘lived experience’, rather than a gruesome image (as in the graphic health effect condition), and also included additional text in the form of a personal testimonial (see Supplementary Figure). It is possible that warnings with personal testimonials take longer to process, and may have greater long-term impacts rather than immediate impacts. Future research should consider examining the potential longer-term impacts of personal testimonials.

Lastly, these findings replicate evidence from high income countries, meaning that if existing graphic images can be used on warnings, development costs may be reduced, as novel images will not need to be created or culturally adapted. This is of utmost importance in developing countries, where financial resources for public health initiatives may be limited.

4.1. Limitations

It is difficult to imply a true mediational relationship, given that the current study is based on cross-sectional data (Winer, Cervone, Bryant, McKinney, Liu, & Nadorff, 2016). However, the temporal order of variables in the proposed model were based on a theoretical framework, including affect heuristic, which posits that emotions, such as fear, guide our cognitive responses and decision-making (Slovic, Finucane, Peters, & MacGregor, 2007). Nonetheless, future research should endeavor to address this issue of temporality, and examine whether these associations persist over time.

With respect to the graphic health effect and personal testimonial warnings, the text and image were different in each of these conditions, thus, it was not possible to distinguish whether the observed effect was due to the image or the additional text in the form of a personal testimonial. Future research should aim to examine the mechanism driving this effect. In terms of the nature of the stimuli, images for the current study were selected and allocated to experimental conditions based on previous research and local consultation (see Supplementary Figure for images and their respective sources). Images were also pre-tested locally prior to conducting the study.

Given the experimental nature of the study, it was not possible to mimic the effects of “real-world” exposure to health warnings, where users see the warnings multiple times over prolonged periods of time on the product, rather than viewing them once on a screen (out of the context of use) and not on a pack for added context. Exposure to health warnings in real-world settings tends to be more passive than in the context of an experimental study. This immediate, forced exposure may have led to stronger reactions to warnings, particularly those with graphic health effects. Given that India and Bangladesh have committed to implementing larger, graphic warnings, research should seek to evaluate their actual effectiveness, over time.

5. CONCLUSION

This set of findings adds to the evidence base in LMICs that graphic, fear-arousing images have potential as an effective tool for health communication within tobacco control. Graphic pictorial health warnings were shown to elicit the greatest levels of negative affect, and in turn, were perceived as most effective. Of particular importance is that these findings mirror patterns found in high-income countries with respect to cigarette warnings, suggesting that individuals’ responses to different types of message content may be similar across diverse cultural environments.

Supplementary Material

Supplemental Figure 1

Footnotes

1

These changes occurred after data collection for the current study (April to August 2012).

2

‘TOBACCO KILLS’ was the text on Indian smokeless tobacco health warning labels at the time of the study.

3

In order to estimate the direct and indirect effects of all k X variables, PROCESS is executed k times, with each run putting one Xi in the model as X and the remaining k-1 X variables as covariates. In addition, the random number generator is seeded with a common (arbitrary) seed (we chose 3,050) so that at each run, the bootstrap confidence intervals were based on the same set of 5,000 resamples from the dataset (Hayes, 2013).

REFERENCES

  1. Adkison SE, Bansal-Travers M, Smith DM, O’Connor RJ, & Hyland AJ (2014). Impact of smokeless tobacco packaging on perceptions and beliefs among youth, young adults, and adults in the U.S: findings from an internet-based cross-sectional survey. Harm Reduction Journal, 11(2). doi: 10.1186/1477-7517-11-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Arora M, Tewari A, Nazar GP, Gupta VK, & Shrivastav R (2012). Ineffective pictorial health warnings on tobacco products: lessons learnt from India. Indian Journal of Public Health, 56(1), 61–4. doi: 10.4103/0019-557X.96978 [DOI] [PubMed] [Google Scholar]
  3. Borland R, Yong H, Wilson N, Fong GT, Hammond D, Cummings KM, … McNeill A (2009). How reactions to cigarette packet health warnings influence quitting: Findings from the ITC Four‐Country survey. Addiction, 104(4), 669–75. doi: 10.1111/j.1360-0443.2009.02508 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. BRC Marketing and Social research. (2004). Smoking health warnings study: Optimizing smoking health warnings - Stage 2 text graphics, size and color testing. Wellington, NZ: Ministry of Health; Retrieved fromhttp://www.tobaccolabels.ca/wp/wp-content/uploads/2013/12/NZ-2004-Optimising-Smoking-Health-Warnings-Stage-2-Text-Graphics-Size-and-Colour-Testing-Government-Report.pdf [Google Scholar]
  5. Byrne S, Katz SJ, Mathios A, & Niederdeppe J (2015). Do the ends justify the means? A test of alternatives to the FDA proposed cigarette warning labels. Health Communication, 30(7), 680–693. doi: 10.1080/10410236.2014.895282 [DOI] [PubMed] [Google Scholar]
  6. Callery WE, Hammond D, O’Connor RJ, & Fong GT (2011). The appeal of smokeless tobacco products among young Canadian smokers: the impact of pictorial health warnings and relative risk messages . Nicotine & Tobacco Research, 13(5), 373–83. doi: 10.1093/ntr/ntr013 [DOI] [PubMed] [Google Scholar]
  7. Cho YJ, Thrasher JF, Swayampakala K, Yong HH, McKeever R, Hammond D, Anshari D, Cummings KM, Borland R (2016) Does reactance against cigarette warning labels matter? Warning label responses and downstream smoking cessation amongst adult smokers in Australia, Canada, Mexico and the United States. PLoS ONE. 11(7): e0159245 10.1371/journal.pone.0159245 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Elliot & Shanahan Research. (2003). Developmental research for new Australian health warnings on tobacco products - Stage 2. Sydney, AU: Population Health Department of Health and Ageing; Retrieved from: http://who.int/fctc/guidelines/ArtElevenAustraliaTwentytwo.pdf [Google Scholar]
  9. Emery LF, Romer D, Sheerin KM, Jamieson KH, & Peters E (2014). Affective and cognitive mediators of the impact of cigarette warning labels. Nicotine & Tobacco Research, 16(3), 263–269. doi: 10.1093/ntr/ntt124 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Environics Research Group. (1999). Canadian adult and youth opinions on sizing of health warning messages. Toronto: Environics Reseach Group Limited. [Google Scholar]
  11. Environics. (2000). Testing new health warning messages for cigarette packages: a summary of three phases of focus group research - Final report. Prepared for Health Canada. Retrieved from: http://www.tobaccolabels.ca/research-reports/hwreports/?n=Canada [Google Scholar]
  12. Evans AT, Peters E, Strasser AA, Emery LF, Sheerin KM, & Romer D (2015). Graphic Warning Labels Elicit Affective and Thoughtful Responses from Smokers: Results of a Randomized Clinical Trial. PloS one, 10(12), e0142879. doi: 10.1371/journal.pone.0142879 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gupta PC, & Ray CS (2003). Smokeless tobacco and health in India and South Asia. Respirology, 8(4), 419–31. doi: 10.1046/j.1440-1843.2003.00507.x [DOI] [PubMed] [Google Scholar]
  14. Gravely S, Fong GT, Driezen P, Xu S, Quah AC, Sansone G, ... & Pednekar MS (2016). An examination of the effectiveness of health warning labels on smokeless tobacco products in four states in India: findings from the TCP India cohort survey. BMC Public Health, 16(1), 1246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hammond D, Fong GT, McDonald PW, Brown KS, & Cameron R (2004). Graphic Canadian cigarette warning labels and adverse outcomes: evidence from Canadian smokers. American Journal of Public Health, 94(8), 1442–5. doi: 10.2105/AJPH.94.8.1442 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hammond D, Thrasher J, Reid JL, Driezen P, Boudreau C, & Santillán EA (2012). Perceived effectiveness of pictorial health warnings among Mexican youth and adults: a population-level intervention with potential to reduce tobacco-related inequities. Cancer Causes & Control, 23(Suppl 1), 57–67. doi: 10.1007/s10552-012-9902-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hammond D (2011). Health warning messages on tobacco products: a review. Tobacco Control, 20(5), 327–37. doi: 10.1136/tc.2010.037630 [DOI] [PubMed] [Google Scholar]
  18. Hayes AF (2013). Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. New York, NY, US: Guilford Press. [Google Scholar]
  19. International Agency for Research on Cancer (IARC) Working Group on the Evaluation of Carcinogenic Risks to Humans. (2007). IARC Monographs on the Evaluation of Carcinogenic Risk to Humans - Volume 89: Smokeless Tobacco and Some Tobacco-specific N-Nitrosamines. Retrieved from http://monographs.iarc.fr/ENG/recentpub/mono89.pdf [PMC free article] [PubMed]
  20. International Institute for Population Sciences (IIPS). (2010). Global Adult Tobacco Survey India (GATS India). New Delhi, India. [Google Scholar]
  21. Kakde S, Bhopal RS, & Jones CM (2012). A systematic review on the social context of smokeless tobacco use in the South Asian population: implications for public health. Public Health, 126(8), 635–45. doi: 10.1016/j.puhe.2012.05.002 [DOI] [PubMed] [Google Scholar]
  22. Kees J, Burton S, Andrews JC, & Kozup J (2010). Understanding how graphic pictorial warnings work on cigarette packaging. Journal of Public Policy & Marketing, 29(2), 265–76. doi: 10.1509/jppm.29.2.265 [DOI] [Google Scholar]
  23. Kessels LTE, Ruiter RAC, Wouters L, & Jansma BM (2014). Neuroscientific evidence for defensive avoidance of fear appeals. International Journal of Psychology, 49(2), 80–8. doi: 10.1002/ijop.12036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kessels LT, Ruiter RA, & Jansma BM (2010). Increased attention but more efficient disengagement: Neuroscientific evidence for defensive processing of threatening health information. Health Psychology, 29(4), 346–54. doi: 10.1037/a0019372 [DOI] [PubMed] [Google Scholar]
  25. Kreuter MW, Green MC, Cappella JN, Slater MD, Wise ME, Storey D, … Woolley S (2007). Narrative communication in cancer prevention and control: a framework to guide research and application. Annals of Behavioral Medicine, 33(3), 221–35. [DOI] [PubMed] [Google Scholar]
  26. Khan ZU (2012). An overview of oral cancer in Indian subcontinent and recommendations to decrease its incidence. Webmed Central CANCER, 3(8). doi: 0.9754/journal.wmc.2012.003626 [Google Scholar]
  27. Khawaja M, Mazahir S, Majeed A, Malik F, Merchant K, Maqsood M, … Fatmi Z (2006). Chewing of betel, areca and tobacco: perceptions and knowledge regarding their role in head and neck cancers in an urban squatter settlement in Pakistan. Asian Pacific Journal of Cancer Prevention, 7(1), 95–100. [PubMed] [Google Scholar]
  28. Larkey LK, & Hecht M (2010). A model of effects of narrative as culture-centric health promotion. Journal of Health Communication, 15(2), 114–35. doi: 10.1080/10810730903528017 [DOI] [PubMed] [Google Scholar]
  29. McElroy A (1990). Biocultural Models in Studies of Human Health and Adaptation. Medical Anthropology Quarterly, 4(3), 243–65. doi: 10.1525/maq.1990.4.3.02a00010 [DOI] [Google Scholar]
  30. Messina J, Freeman C, Rees A, Goyder E, Hoy A, Ellis S, & Ainsworth N (2013). A systematic review of contextual factors relating to smokeless tobacco use among South Asian users in England. Nicotine & Tobacco Research, 15(5), 875–82. doi: 10.1093/ntr/nts193 [DOI] [PubMed] [Google Scholar]
  31. Mutti S, Reid JL, Gupta PC, Pednekar MS, Dhumal G, Nargis N, … Hammond D (2015). Perceived effectiveness of text and pictorial health warnings for smokeless tobacco packages in Navi Mumbai, India, and Dhaka, Bangladesh: findings from an experimental study. Tobacco Control. doi: 10.1136/tobaccocontrol-2015-052315 [DOI] [PubMed] [Google Scholar]
  32. National Cancer Institute and Centers for Disease Control and Prevention. Smokeless Tobacco and Public Health: A Global Perspective. Bethesda, MD: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention and National Institutes of Health, National Cancer Institute; NIH Publication No. 14–7983; 2014. Retrieved from: http://cancercontrol.cancer.gov/brp/tcrb/global-perspective/index.html. [Google Scholar]
  33. Netemeyer RG, Burton S, Andrews JC, & Kees J (2016). Graphic Health Warnings on Cigarette Packages: The Role of Emotions in Affecting Adolescent Smoking Consideration and Secondhand Smoke Beliefs. Journal of Public Policy & Marketing, 35(1), 124–143. doi: 10.1509/jppm.15.008 [DOI] [Google Scholar]
  34. Noar SM, Hall MG, Francis DB, Ribisl KM, Pepper JK, & Brewer NT (2015). Pictorial cigarette pack warnings: a meta-analysis of experimental studies. Tobacco Control. doi: 10.1136/tobaccocontrol-2014-051978 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Oswal KC, Raute LJ, Pednekar MS, & Gupta PC (2011). Are current pictorial warnings in India effective? Asian Pacific Journal of Cancer Prevention, 12, 121–4. [PubMed] [Google Scholar]
  36. Pepper JK, Cameron LD, Reiter PL, McRee AL, & Brewer NT (2013). Non-smoking male adolescents’ reactions to cigarette warnings. PloS one, 8(8), e65533 10.1371/journal.pone.0065533 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Peters E, Romer D, Slovic P, Jamieson KH, Wharfield L, Mertz CK, & Carpenter SM (2007). The impact and acceptability of Canadian-style cigarette warning labels among U.S. smokers and nonsmokers. Nicotine & Tobacco Research, 9(4), 473–81. doi: 10.1080/14622200701239639 [DOI] [PubMed] [Google Scholar]
  38. Petty RE, Cacioppo JT The elaboration likelihood model of persuasion. Springer; New York; 1986. [Google Scholar]
  39. Pierce JP, Choi WS, Gilpin EA, Farkas AJ, & Merritt RK (1996). Validation of susceptibility as a predictor of which adolescents take up smoking in the United States. Health psychology, 15(5), 355–361. 10.1037/0278-6133.15.5.355 [DOI] [PubMed] [Google Scholar]
  40. Preacher KJ & Hayes AF (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36(4), 717–31. doi: 10.3758/BF03206553 [DOI] [PubMed] [Google Scholar]
  41. Rahman MA, Mahmood MA, Spurrier N, Rahman M, Choudhury SR, & Leeder S (2012). Why do Bangladeshi people use smokeless tobacco products? Asia-Pacific Journal of Public Health, 27(2), 197–209. doi: 10.1177/1010539512446957 [DOI] [PubMed] [Google Scholar]
  42. Slovic P, Finucane ML, Peters E, & MacGregor DG (2007). The affect heuristic. European Journal of Operational Research, 177(3), 1333–1352. 10.1016/j.ejor.2005.04.006 [DOI] [Google Scholar]
  43. Strahan EJ, White K, Fong GT, Fabrigar LR, Zanna MP, Cameron R (2002). Enhancing the effectiveness of tobacco package warning labels: a social psychological perspective. Tobacco Control, 11(3):183–90. doi: 10.1136/tc.11.3.183 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Sudman S (1980). Improving the quality of shopping center sampling. Journal of Marketing Research, 17(4), 423–31. doi: 10.2307/3150496 [DOI] [Google Scholar]
  45. Thrasher JF, Swayampakala K, Hammond D, Bansal-Travers M, Hardin J, Yong H, … Borland R (2016). Influences of self-efficacy, response efficacy and reactance on responses to cigarette health warnings: A longitudinal study of adult smokers in Australia and Canada. Health Communication. 31(12): 1517–1526. doi: 10.1080/10410236.2015.1089456 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Thrasher JF, Arillo-Santillán E, Villalobos V, Pérez-Hernández R, Hammond D, Carter J, … Regalado-Piñeda J (2012). Can pictorial warning labels on cigarette packages address smoking-related health disparities? Field experiments in Mexico to assess pictorial warning label content. Cancer Causes & Control, 23(1), 69–80. doi: 10.1007/s10552-012-9899-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wang AL, Romer D, Elman I, Turetsky BI, Gur RC, & Langleben DD (2015). Emotional graphic cigarette warning labels reduce the electrophysiological brain response to smoking cues. Addiction biology, 20(2), 368–376. doi: 10.1111/adb.12117 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Witte K (1992). Putting the fear back into fear appeals. The extended parallel processing model. Communication Monographs, 59(4), 329–349. doi: 10.1080/03637759209376276 [DOI] [Google Scholar]
  49. Witte K, & Allen M (2000). A meta-analysis of fear appeals: Implications for effective public health campaigns. Health, Education & Behavior, 27(5), 591–615. doi: 10.1177/109019810002700506 [DOI] [PubMed] [Google Scholar]
  50. Winer ES, Cervone D, Bryant J, McKinney C, Liu RT, & Nadorff MR (2016). Distinguishing mediational models and analyses in clinical psychology: atemporal associations do not imply causation. Journal of Clinical Psychology, 72(9), 947–955. doi: 10.1002/jclp.22298 [DOI] [PubMed] [Google Scholar]
  51. World Health Organization. (2008). Guidelines for implementation of Article 11 of the WHO framework convention on tobacco. Retrieved from www.who.int/fctc/guidelines/article_11.pdf
  52. World Health Organization. (2009). Global Adult Tobacco Survey (GATS) Bangladesh Report 2009. Dhaka, Bangladesh: Ministry of Health and Family Welfare. [Google Scholar]
  53. Yong HH, Borland R, Thrasher JF, Thompson ME, Nagelhout GE, Fong GT, ... & Cummings KM (2014). Mediational pathways of the impact of cigarette warning labels on quit attempts. Health Psychology, 33(11), 1410. doi: 10.1037/hea0000056 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Figure 1
NIHMS985686-supplement-Supplemental_Figure_1.docx (584.3KB, docx)

RESOURCES