Yoo Jin Cho

Date of Award

Fall 2020

Document Type

Open Access Dissertation


Health Services and Policy Management

First Advisor

James F. Thrasher


Package warning labeling policy is a fundamental public health strategy for communicating about tobacco product risks. The rapidly growing markets for novel tobacco products, such as heated tobacco products (HTPs) and e-cigarettes (ECs), make it critical to determine the characteristics of effective warning label policies for these products, which have lower levels of risk compared to cigarettes according to available evidence. Whether warning labels can communicate information about a continuum of risk across tobacco products is unexplored.

This dissertation investigated two warning label systems for communicating the relative risks of cigarettes, HTPs, and ECs: (1) varying warning sizes as a heuristic for differential risk, whereby warnings occupy different percentages of the packages for each product type; (2) “dashboard” icons (e.g., as used for fire risk) integrated into warnings, with colors used to indicate relative risk of each product type. We used an alternative specific discrete choice experiment (DCE) that included a 2X2 between-subject manipulation (i.e., different vs. equal size; presence vs absence of dashboard), as well as a within-subject evaluation of 2 (brand) x 2 (warning image type) x 3 (product type) x 4 (product flavor) effects on participant perceptions and behavioral intentions. This experimental design resulted in 160 different product packages allocated to 32 different choice sets, each of which contained one cigarette, one EC, and one HTP package. To reduce response burden, participants were randomized to one of two blocks of 16 choice sets. In 2019, 1,280 Korean adults were recruited from an online commercial panel, including 444 young adult non-users of cigarettes, HTPs, or ECs and 836 current users of cigarettes, HTPs, or ECs. After reading descriptions of each product, participants completed the DCE, in which they evaluated 16 sets of three product alternatives. In each choice set, participants chose the packages that most and least elicited positive affect and the products that most and least aroused interest in trying. They could also “opt out” to indicate that the packages or products were no different from each other.

In the first manuscript, we estimated logistic regression models to examine the main and interaction effects of the between-subject manipulations on perceptions of relative harm and benefits of switching, which were assessed after completing the DCE. Exposure to the dashboard icon was associated with reporting higher perceived harm for cigarettes than ECs, cigarettes than HTPs, and HTPs than ECs, as well as perceived reduced harm of switching from cigarettes to HTPs, cigarettes to ECs, and HTPs to ECs. No main effects were found for the different-size warning label system; no interaction with the dashboard icon was found.

In the second manuscript, we estimated conditional logistic choice models to examine the effects of the warning systems on choices around positive affect and interest in trying in the context of the DCE. Exposure to either the dashboard icon, different-size warning system or both systems significantly decreased the relative positive affect towards cigarettes compared to ECs and HTPs compared to ECs, as well as the relative interest in trying cigarettes compared to ECs and HTPs compared to ECs.

We conclude that dashboard icons with colors are a potential strategy for representing the continuum of risks to enhance consumer understanding about the relative risks across tobacco products. Varying warning sizes across different product types may increase the dashboard icon effects by eliciting more positive emotional responses and promoting consumer intentions to choose less risky products.

Available for download on Sunday, May 08, 2022