Date of Award


Document Type

Open Access Dissertation


Health Promotion, Education and Behavior


Norman J. Arnold School of Public Health

First Advisor

James F. Thrasher


Pictorial health warning labels (PHWL) on cigarette packaging is a key way to communicate with consumers about the harms from tobacco, particularly in the low- and middle-income countries that do not have the resources for effective mass media campaigns. Research is needed to determine the most effective PHWL content for Indonesia, a country with one of the largest populations of smokers in the world and amongst the weakest tobacco policy environments. This research aimed to determine the most effective PHWL content for Indonesia’s cigarette packages, including the social and psychological factors that may influence PHWL effects.

Data for this study came from a field experiment with Indonesian adult smokers (n=584), and 15- to 18-year-old adolescent smokers (n=280) and nonsmokers (n=313) using both between- and within-subject manipulations. First, we assessed effects of health warning label (HWL) characteristics, including warning type (text-only versus pictorial warnings = within subject), imagery type (graphic, suffering, and symbolic imagery = within subject), and textual type (didactic versus testimonial = between subject), on negative emotional responses, message credibility, and perceived effectiveness of the HWLs. Second, we assessed whether the effects of HWLs on these outcomes were moderated by variables for which theories indicated differential responses to HWLs were likely (i.e., smoker identity and self-efficacy to quit among smokers, reactance to HWL stimuli and advertising exposure among all participants). Main and interactive effects of HWL manipulations and participant characteristics on outcomes were estimated using linear mixed effects models to adjust for correlated data due to repeated measures.

We found that compared to text-only warnings, PHWLs were rated significantly higher on all outcomes. Within the PHWLs, those with graphic imagery were rated the highest on all outcomes, followed by suffering imagery, and symbolic imagery. No significant differences were found between textual types for any outcome. Smoker identity was negatively associated only with perceived effectiveness, with no significant interactions found. Self-efficacy was positively associated with all outcomes, finding a significant interaction with imagery type in models for negative emotions, suggesting that rating differences between text-only HWLs and symbolic PHWLs were greater amongst those with higher self-efficacy. Reactance was positively associated with all outcomes, significantly interacting with imagery type in models assessing negative emotions and perceived effectiveness. This suggests the differences between symbolic and suffering PHWLs were greater amongst those with low reactance than those with high reactance, although the pattern of results with regard to which HWL image styles had the strongest effects was the same. Advertising exposure was positively associated with all outcomes and significantly interacting with textual and imagery types when assessing message credibility and perceived effectiveness. This suggests that didactic HWLs were rated lower than testimonials in low exposure group but were rated higher in high exposure group, while differences between graphic and suffering PHWLs were greater in low exposure than in high exposure groups.

Overall, specific types of HWL content produced a pattern of responses for Indonesia that is similar to other countries. Our findings add further support for FCTC recommendations to adopt graphic PHWLs, with no evidence found to suggest the negative effects for PHWLs in key subpopulations.


© 2017, Dien Anshari