Effects of Pictorial Warning Label Message Framing and Standardized Packaging on Cigarette Packaging Appeal Among Young Adult Smokers

Abstract

Introduction:

Pictorial warning labels and standardized “plain” packaging are policy interventions to reduce smoking, in part, by making cigarette packs and smoking less appealing. To inform potential policy decisions, this study examined the relative effects of message text framing (gain vs. loss) and cigarette packaging (standardized vs. branded) on appeal in a sample of young adult cigarette smokers.

Methods:

Cigarette smokers (N=339) ages 18-30 completed two within-subjects experimental tasks. Tasks assessed the effects of message text framing (gain vs. loss) and packaging (standardized vs. branded) on cigarette packaging appeal. Task 1 was a 2x2 discrete choice experiment, where participants chose between each experimental pack and a standard branded cigarette pack without a pictorial warning label. Task 2 was a ranking task where participants ranked all packs on measures of appeal.

Results:

In Task 1, there were no significant differences in measures of appeal between packs displaying gain- vs. loss-framed message text, but all packs with pictorial warning labels significantly decreased appeal relative to standard branded packs without pictorial warning labels. Standardized packs with pictorial warning labels significantly reduced appeal relative to branded packs with pictorial warning labels and standard branded packs without pictorial warning labels. Task 2 pack rankings showed similar effects of pictorial warning labels and standardized packaging on appeal.

Conclusions:

Pictorial warning labels with gain- and loss-framed text were equally powerful at reducing appeal of cigarette packs in young adult smokers relative to branded packs without pictorial warning labels, especially when combined with standardized packaging.

INTRODUCTION

Cigarette advertising and marketing increases initiation and escalation of cigarette smoking in young people.¹ Cigarette advertising and marketing appeals to consumers through branding and carefully designed, population-targeted features.² As countries globally restricted cigarette advertising and marketing in order to curb the appeal it creates, cigarette companies turned to packaging as a marketing tool.^1,2 Consequently, cigarette packaging regulations such as pictorial warning labels (PWLs) communicating the risks of smoking on cigarette packs are recommended as a tobacco control policy intervention .^3,4

The Framework Convention on Tobacco Control recommends PWLs covering a minimum of 30% of the surface area of the front and back of cigarette packages.³ More than 100 countries have adopted PWLs for cigarette packages,^5,6 and their adoption has contributed to reductions in smoking and smoking-attributable morbidity and mortality.^7-10 PWLs deter smoking uptake and promote quitting through multiple processes, such as capturing attention, eliciting emotional response, increasing encoding and perceived message effectiveness, affecting smoking-related beliefs, and increasing motivation to avoid or quit smoking.^8,9,11-13

Despite widespread global adoption, the United States has not implemented PWLs on cigarette packages. The 2009 Family Smoking Prevention and Tobacco Control Act, which authorized the Food and Drug Administration (FDA) to regulate tobacco products, required PWLs for cigarette packs.¹⁴ However, a 2011 rule issued by FDA to implement this requirement was withdrawn in response to lawsuits by tobacco companies.^15,16 In March 2020, FDA issued a new rule with updated PWLs to be required in the U.S.¹⁷ The Tobacco Control Act also authorizes FDA to change required PWLs in the future (e.g., update text, images), however any future steps will be contingent on initial PWLs implementation.

“Plain” or standardized cigarette packaging (herein, standardized packaging) is another policy intervention with increasing global adoption.¹⁸ Standardized packaging removes industry branding from cigarette packs by requiring packs to be a uniform color (e.g., brown) to reduce the appeal of cigarette packaging.^19-22 Evidence demonstrates standardized packaging contributes to reductions in smoking (e.g., in Canada, Australia).^23-26 The FDA could conceivably implement standardized packaging in the U.S. if evidence continues to support its public health impact, though legal challenges will inevitably arise and strong evidence will be required.¹⁴

For PWLs, it is critical for research to identify the PWL content that will lead to the greatest reductions in smoking. For countries yet to adopt PWLs, this evidence can inform policy development. After countries have implemented PWLs, this evidence can inform decisions on how to rotate PWL messages or update them over time to minimize potential “burnout” (i.e., weakened effects as people see PWLs repeatedly).^27,28 Studies can also provide evidence on the effects of PWL messaging in settings with different packaging policies. Some PWL messaging strategies may be optimal in contexts where branded cigarette packaging is present, such as the U.S.²²

According to Prospect Theory,²⁹ PWL messages can be framed around gains of quitting smoking (i.e., the health benefits of quitting) or losses of continuing to smoke (i.e., the health risks of smoking). The prevailing strategy globally and proposed in the U.S. for PWLs is loss-framed messages communicating the risks of smoking (e.g., smoking causes cancer).¹⁷ However, one experimental study examining the effects of gain- and loss-framed PWL message text among U.S. young adult smokers demonstrated that warnings with gain-framed text and pictorial depictions of smoking risks may be effective to diversify PWL messages without reducing impact.³⁰ Further, this study showed that in contexts with standardized packaging, gain-framed PWL text may be optimal.³⁰ Research examining the effects of PWL message features such framing relative to packaging policies is important to inform policy decisions in the U.S. and globally.

This study extends the evidence on PWL message framing and standardized packaging in important ways. Prior research on PWL message framing focused on motivation to quit smoking as an outcome.³⁰ We add to this research by investigating the effects of PWL message framing (gain- vs. loss-framed) and packaging (standardized vs. branded) on experimental measures of cigarette packaging appeal. Behavioral experimental methods are valuable for investigating the appeal of cigarette packaging features and examining how policies can reduce appeal.^31,32 Reducing packaging appeal is an important pathway through which PWLs and standardized packaging affect smoking behavior,¹⁹ but to our knowledge there is no existing evidence on the impact of PWL message framing on cigarette packaging appeal in contexts with standardized and branded packaging. Like prior studies,³⁰ our research focused on young adult smokers because this is a population where smoking behavior solidifies,³³ where the long-term benefits of quitting smoking are maximized,³⁴ and where there is an identified need for tobacco control policy research.³⁵ We hypothesized that compared with branded packaging with text-only warnings, PWLs and standardized packaging will reduce the appeal of cigarettes among young adult cigarette smokers, and PWLs with gain-framed messages on standardized packages will have the strongest effects on reducing appeal.

METHODS

Procedures

We collected data for this study from December 2014 to September 2019. Participants were young adult cigarette smokers recruited from the community in a major metropolitan area in the U.S. mid-Atlantic region. For recruitment, we used flyers in the community, social media and newspaper advertisements, online classifieds, recontacted participants from prior non-treatment studies, and received direct referrals. Eligible participants were: (1) age 18-30 years; (2) smoked ≥100 lifetime cigarettes; (3) currently smoked cigarettes every day or some days, (4) willing to send and receive text messages via a personal mobile phone for the prospective portion of the study (not reported here); and (5) able to complete study procedures in English. The cigarette smoking eligibility criteria followed standard epidemiologic definitions of current smoking in the U.S. population.³⁶ We assessed eligibility by telephone and participants signed informed consent forms. The host institution’s institutional review board approved study procedures, and the study was registered in ClinicalTrials.gov (NCT03446170).

We sent eligible, consenting participants a link to a secure, online survey with two-embedded experimental tasks. Before the experimental tasks, participants completed measures of demographics, smoking history, other tobacco use, and smoking-related beliefs and perceptions. We designed the tasks based on recommended behavioral experimental methods^31,32 to assess participants’ choices and preferences as measures of appeal when presented with cigarette packs with varying attributes. We used two tasks to capture unique measures of participants’ choices and their ranking preferences for packages with experimentally manipulated attributes as distinct measures of appeal.³⁷

The attributes we tested in both tasks were PWL message framing (gain vs. loss) and packaging (standardized vs. branded). We adapted stimuli from a prior investigation.³⁸ For the PWL attribute, we used a PWL communicating the risks of cancer from smoking from the warnings FDA issued in 2011.¹⁵ The loss-framed PWL message text was unmodified: “WARNING: Cigarettes cause cancer.” We modified text for the gain-frame PWL to “WARNING: Quitting smoking reduces the risk of cancer.” The image for both gain- and loss-framed warnings depicted mouth cancer. We chose this warning because a prior study confirmed the framing manipulation and indicated it had the strongest effect on motivation to quit in young adult smokers in a prior study.^30,38,39

For the packaging attribute, we used an unfamiliar cigarette brand (“Peter Jackson”) to account for the potential impact of smokers’ brand preferences in the experimental design.²⁶ The branded pack images displayed gain- and loss-framed PWLs on packs with unaltered branding (i.e., the manufacturer’s typography, color, imagery). The standardized packages were a uniform drab brown color based on those implemented in other countries globally.²² They listed the brand name in black, sans serif font on the front of the pack.³

For the experimental tasks, we created three-dimensional images with dimensions sized to those of a standard U.S. cigarette pack and PWLs covering 50% of the pack surface area based on the Tobacco Control Act’s requirements.¹⁴ The experimental stimuli are included Supplementary Figure 1 and 2.

Task 1 was a 2x2 within-subjects discrete choice experiment⁴⁰ where we tested two attributes: PWL message framing (gain vs. loss) and cigarette packaging (standardized vs. branded). Because the number of attributes and the levels within each attribute was relatively small, we used a complete factorial design rather than a reduced design where participants viewed a subset of all possible choice sets.⁴¹ Participants viewed 4 choice sets (i.e., one set for each pack variation in the 2x2 design). In each choice set, participants chose between one of the experimental packs and a standard branded cigarette pack with no PWL reflecting the cigarette packaging and labeling requirements in the U.S. at the time of the study. We randomized the order of the choice sets and the order in which the packs appeared on the screen within each set within-subjects. For each choice set, we instructed participants to choose from two packs in response to five outcome measures, detailed below. This task enabled us to assess the independent effects of each attribute and their interactions on participants’ choices in relation to a branded cigarette package with no PWL.

Task 2 was a ranking task⁴² where we also tested the effects of the PWL message framing (gain vs. loss) and cigarette packaging (standardized vs. branding) attributes on participants’ preferences based on their rankings of the packs. We displayed the 4 experimental packs and the standard branded cigarette pack on the same screen, with the order of appearance across the screen randomized. We asked participants to rank the packs on 5 measures of appeal, detailed below. Rankings were mutually exclusive such that a ranking of 1, 2, 3, 4, or 5 could be provided for only 1 pack. This task enabled us to assess the effects of each attribute in relation to a branded cigarette package with no PWL and relative to each other in rank order.

Measures

Demographics.

We assessed demographic characteristics including age, sex, race, ethnicity, education, and employment.

Cigarette Smoking.

We assessed age of smoking initiation, cigarettes smoked per day and nicotine dependence using the Fagerstrom Test for Nicotine Dependence (FTND).⁴³

Other Tobacco Use.

We measured other tobacco use in the past 30 days including cigars/little cigars/cigarillos, electronic cigarettes, waterpipe/hookah, and smokeless tobacco.⁴⁴

Experimental Tasks.

For both tasks, we selected outcome measures from prior research.^13,31,45 For Task 1, for each choice set we asked participants to choose which pack: 1. They would be most likely to buy, 2. Is most appealing to them, 3. Attracts their attention, 4. Contains cigarettes that are most harmful to their health, and 5. Makes them think about the health risks of smoking. We analyzed participants’ choices for each outcome separately.

For Task 2, participants viewed all 5 packs and ranked them based on which: 1. They would be most likely to buy, 2. Is most appealing to them, 3. Is most attractive, 4. Contains cigarettes that are most harmful to their health, and 5. Makes them think about the health risks of smoking, and 6. Makes them want to quit smoking. Ranks ranged from 1 (Least) to 5 (Most). We analyzed participants’ rankings for each item separately, and we created an average rank variable where higher values indicate greater appeal by reverse-coding items 4, 5, and 6.

Analysis

We used descriptive statistics to characterize the sample. To analyze outcomes, we used multivariable models with generalized estimating equations (GEE) with an exchangeable working correlation structure to account for the repeated measures design. We chose this analytic approach because it enabled us to test study hypotheses, has been used in similar experimental tobacco policy research, and was appropriate for the experimental design.⁴⁶ For Task 1, we coded outcomes dichotomously (choice=1, no choice=0) to reflect participants’ choices of the experimental packs versus the standard branded pack. Using logistic regression models with GEE, we tested the effects for PWL message framing (0 = loss, 1 = gain), standardized packaging (0 = standardized, 1 = branded), and their interaction in separate models for each outcome. The interaction effect was not significant for any outcome, so we removed it from the models reported. We also tested if results with and without an interaction differed with covariates (cigarettes per day, age, gender, race, ethnicity, education, and study year). Covariates worsened model fit (lower values of Quasi-likelihood under the Independence model Criterion (QIC)) and parameter estimates did not substantially differ with covariates, so we report unadjusted models. We report results as odds ratios with 95% confidence intervals.

For Task 2, we modeled the rankings for each outcome variable using linear models with GEE with a categorical variable for the 5 packs as the only independent variable. Like the analyses for Task 1, covariates did not affect the model results and worsened model fit, so report unadjusted models. We present estimated coefficients and 95% confidence intervals and least squared mean differences in using a Tukey-Kramer adjustment for multiple testing. Overall, there were minimal missing data (no more than 3% for any given variable), so we analyzed the data with listwise deletion. We used SAS version 9.4 for all analyses.

RESULTS

Participants.

In total, 796 individuals responded to recruitment materials and 647 (81%) were eligible. The most common reasons for ineligibility were being outside the study age range (59, 40% of those ineligible) and not being current smokers (59, 40% of those ineligible). A total of 365 eligible individuals provided informed consent (56% of those eligible), and 339 eligible, consenting participants (93%) completed the study procedures.

Table 1 shows characteristics of the sample. Participants were demographically diverse (46% female, 54% non-white race, 9% Hispanic ethnicity, 56% annual income $35,000 or less), averaging 24.8 (SD 3.4) years of age. Participants initiated smoking at 17.2 (SD 2.7) years of age on average, smoked 8.1 (SD 6.8) cigarettes per day on average, and reported a low mean FTND dependence score 2.7 (SD 2.4). Overall, 27% used electronic cigarettes, 23% hookah (23%), and 30% cigars/little cigars (30%) in the past 30 days.

Table 1.

Sample Characteristics (N=339)

	M (SD)	n (%)
Age	24.8 (3.4)	-
Gender
Male	-	181 (53%)
Female	-	157 (46%)
Race
Black/African American	-	109 (32%)
White	-	155 (46%)
Asian	-	41 (12%)
Other	-	31 (9%)
Hispanic
Hispanic	-	30 (9%)
Non-Hispanic	-	308 (91%)
Education
High school or less	-	55 (16%)
Some College	-	144 (42%)
College or more	-	139 (41%)
Annual Income
$35,000 or less	-	189 (56%)
>$35,001	-	148 (44%)
Employment
Full Time Employed	-	226 (67%)
Not Full Time Employed	-	111 (33%)
Age of Smoking Initiation (Years)	17.2 (2.7)	-
Cigarettes per day	8.1 (6.8)	-
Dependence FTND Score (0-10)	2.7 (2.4)	-
Past 30-Day Use of Other Tobacco
Electronic Cigarettes	-	91 (27%)
Waterpipe/hookah	-	78 (23%)
Smokeless Tobacco	-	27 (8%)
Cigars/Little Cigars/Cigarillos	-	100 (30%)

Note. SD = Standard Deviation. Some sample sizes for categories within variables do not sum to total sample size due to sporadic missing data (<3% of cases for any individual variable).

Discrete Choice Task.

Results of the discrete choice task are shown in Table 2. There were no significant differences between packs with gain-framed PWLs and loss-framed PWLs on measures of appeal (Table 2). However, compared to packs with no PWL, packs with gain- and loss-framed PWLs significantly reduced the odds of participants indicating they would buy the pack (OR=0.09, 95% CI: 0.06, 0.12; OR=0.09, 95% CI: 0.07, 0.13) and the pack appealed to them (OR=0.13, 95% CI: 0.09, 0.17; OR=0.11, 95% CI: 0.08, 0.15). Packs with gain- and loss-framed PWLs significantly increased the odds of participants indicating the pack attracted attention (OR=2.77, 95% CI: 2.24, 3.43; OR=2.55, 95% CI: 2.06, 3.16), contains cigarettes that are the most harmful (OR=3.20, 95% CI: 2.54, 4.00; OR=3.17, 95% CI: 2.53, 4.00) and makes them think about risks (OR=9.85, 95% CI: 7.13, 13.59; OR=9.29, 95% CI: 6.78, 12.72) compared to packs with no PWLs (Table 2).

Table 2.

Task 1 Discrete Choice Task Results

Outcome	Gain-Framed PWL vs. Loss-Framed PWL	Gain-Framed PWL vs. No PWL	Loss-Framed PWL vs. No PWL	Standardized Pack, PWL vs. Branded Pack, No PWL	Standardized Pack, PWL vs. Branded Pack, No PWL	Branded Pack, PWL vs. Branded Pack, No PWL
Most likely to buy	0.95 (0.77, 1.18)	0.09 (0.06, 0.12)***	0.09 (0.07, 0.13)***	0.70 (0.51, 0.97)*	0.08 (0.05, 0.11)***	0.11 (0.08, 0.15)***
Is most appealing	1.16 (0.93, 1.44)	0.13 (0.09, 0.17)***	0.11 (0.08, 0.15)***	0.71 (0.56, 0.92)**	0.10 (0.07, 0.14)***	0.14 (0.10, 0.19)***
Attracts attention	1.09 (0.94, 1.26)	2.77 (2.24, 3.43)***	2.55 (2.06, 3.16)***	0.72 (0.61, 0.85)***	2.26 (1.82, 2.79)***	3.14 (2.51, 3.93)***
Contains cigs. that are most harmful	1.00 (0.88, 1.14)	3.20 (2.54, 4.00)***	3.17 (2.53, 4.00)***	1.04 (0.88, 1.23)	3.24 (2.56, 4.10)***	3.12 (2.47, 3.94)***
Makes them think about risks	1.06 (0.84, 1.34)	9.85 (7.13, 13.59)***	9.29 (6.78, 12.72)***	1.12 (0.86, 1.47)	10.15 (7.32, 14.08)***	9.01 (6.52, 12.45)***

Note: Odds Ratios (95% Confidence Intervals)

^*p<.05.

^**p<.01.

^***p<.001.

Pack x Frame interactions were not significant, so they were not included in the final logistic regression models with GEE.

Compared to branded packs with PWLs, standardized packs with PWLs reduced the odds that participants indicated they would buy the pack (OR=0.70, 95% CI: 0.51, 0.97), the pack appealed to them (OR=0.71, 95% CI: 0.56, 0.92), and that it attracted their attention (OR=0.72, 95% CI: 0.61, 0.85). Compared to branded packs with no PWL, standardized packs with PWLs significantly reduced the odds of participants indicating they would buy the pack (OR=0.08, 95% CI: 0.05, 0.11), that the pack appealed to them (OR=0.10, 95% CI: 0.07, 0.14), and significantly increased the odds of participants indicating the pack attracted attention (OR=2.26, 95% CI: 1.82, 2.79), contains cigarettes that are the most harmful (OR=3.24, 95% CI: 2.56, 4.10), and made them think about risks (OR=10.15, 95% CI: 7.32, 14.08) (Table 2). Patterns of findings were similar for branded packs with a PWL compared to branded packs with no PWL (Table 2).

Pack Ranking Task.

Results of the ranking task are shown in Table 3 and Supplementary Figure 3. The branded pack with no PWL produced significantly greater appeal (p < .05) compared to all other packs displaying PWLs on standardized or branded packaging (Table 3). Standardized packs with PWLs were ranked lower on all measures of appeal than branded packs with no PWLs and branded packs with PWLs. For standardized packs displaying PWLs, participants indicated those with loss-framed PWLs were more likely to contain cigarettes that are most harmful (M=3.5 vs. M=3.3), but there were no other differences between standardized packs with gain- and standardized packs with loss-framed PWLs (Table 3). For branded packs displaying PWLs, participants indicated those with loss-framed PWLs were more likely to contain cigarettes that are most harmful (M=3.1 vs. M=2.9), make them think about risks (M=3.2 vs. M=2.9), and make them think about quitting (M=3.2 vs. M=2.9). Compared to branded packs with gain-framed PWLs, participants ranked branded packs with loss-framed PWLs significantly lower on average appeal overall (Table 3).

Table 3.

Task 2 Pack Ranking Task Results

	Branded Control (A)	Branded Loss (B)	Branded Gain (C)	Standardized Loss (D)	Standardized Gain (E)
Most likely to buy	4.2 (4.0, 4.3)B,C,D,E	3.0 (2.9, 3.1)A,C,D,E	3.2 (3.1, 3.3)A,B,D,E	2.4 (2.2, 2.5)A,B,C	2.3 (2.2, 2.5)A,B,C
Is most appealing	4.1 (3.9, 4.3)B,C,D,E	3.0 (2.9, 3.1)A,C,D,E	3.2 (3.1, 3.3)A,B,D,E	2.3 (2.2, 2.5)A,B,C	2.3 (2.2, 2.4)A,B,C
Attracts attention	4.2 (4.1, 4.4)B,C,D,E	3.0 (2.9, 3.1)A,C,D,E	3.2 (3.1, 3.3)A,B,D,E	2.3 (2.1, 2.4)A,B,C	2.3 (2.1, 2.4)A,B,C
Contains cigs. that are most harmful	2.1 (2.0, 2.3)B,C,D,E	3.1 (3.1, 3.2)A,C,D	2.9 (2.8, 3.0)A,B,D,E	3.5 (3.4, 3.7)A,B,C,E	3.3 (3.2, 3.5)A,C,D
Makes them think about risks	1.8 (1.6, 1.9)B,C,D,E	3.2 (3.1, 3.3)A,C,D,E	2.9 (2.8, 3.0)A,B,D,E	3.6 (3.5, 3.8)A,B,C	3.5 (3.4, 3.6)A,B,C
Makes them think about quitting	1.8 (1.7, 2.0)B,C,D,E	3.2 (3.1, 3.3)A,C,D	2.9 (2.8, 3.1)A,B,D,E	3.6 (3.4, 3.7)A,B,C	3.5 (3.3, 3.6)A,C
Average appeal	4.1 (4.0, 4.3)B,C,D,E	2.9 (2.9, 3.0)A,C,D,E	3.1 (3.1, 3.2)A,B,D,E	2.4 (2.3, 2.5)A,B,C	2.4 (2.3, 2.5)A,B,C

Note: Group Mean (Confidence Interval) rankings on a 1 to 5 scale; For average appeal, individual items were re-coded for consistent directionality such that higher scores indicate greater appeal.

DISCUSSION

This study experimentally investigated the effects of gain- vs. loss-framed PWLs and standardized vs. branded packaging on cigarette packaging appeal. This builds on prior studies by demonstrating the effect of PWL framing and standardized packaging on cigarette pack appeal, a precursor to cigarette use, among young adult smokers. In Task 1, there were no significant differences in measures of appeal between gain- and loss-framed PWLs. However, all packs with PWLs significantly decreased appeal relative to branded packs without PWLs. Furthermore, standardized packs with PWLs significantly reduced appeal relative to branded packs with and without PWLs. Patterns of findings from the pack rankings showed similar effects of PWLs and standardized packaging. Yet, for branded packs displaying PWLs, those with loss-framed PWLs were more effective at communicating risks of smoking than gain-framed PWLs.

The findings extend the evidence for PWLs and standardized packaging in several ways. We hypothesized that gain-framed PWLs may have stronger effects than loss-framed PWLs in standardized packaging conditions based on prior research.³⁰ Our results did not show such an interaction, which could be because we examined a different outcome than previous work on which this hypothesis is based. Nevertheless, our results indicate that in countries without standardized packaging regulations such as the U.S., both PWL frames reduce the appeal of cigarette packs. The results also suggest the effect on appeal may be stronger for loss-framed PWLs in such settings. Reducing the appeal of cigarette packaging is an important mechanism of PWL’s effects.^8,9,11,12 The population-level impact of implementing PWLs on outcomes such as reducing smoking-attributable deaths in countries such as the U.S. is evidenced in modeling.⁴⁷ These results provide support for warnings such as those in the FDA’s recently published rule requiring PWLs in the U.S.¹⁷

In both tasks standardized packaging reduced measures of cigarette packaging appeal regardless of PWL message framing. Other research complements these findings by demonstrating an increase in outcomes such as attention to PWLs with standardized packaging.^19-22 Our results indicate loss-framed PWLs were most effective for lowering appeal on branded packs. Yet, standardized packaging stimuli in and of itself produced the greatest reductions in appeal overall regardless of PWL message framing. In countries with standardized packaging policies, the results indicate both gain- and loss-framed PWLs could be effective to reduce smoking among young adults. Interestingly, our results did not indicate that standardized packaging influenced measures of appeal such as making participants think about risks. This is somewhat surprising because standardize packaging is known to increase attention to PWLs, and thereby enhance the salience of risk information.²² However, this result could be due to our relatively broad measure of thinking about risks⁴⁸ or because of the relatively brief exposure. These topics warrant further study.

The findings should be interpreted in light of study limitations. The study was cross-sectional and did not test effects on behaviors. Longitudinal studies should examine this as well as other relevant outcomes (e.g., uptake) for populations not studied here (e.g., youth, nonsmokers). The convenience sample of low nicotine dependent young adult smokers may reduce generalizability. We did not study youth or non-smokers, which are important groups to study in the future to determine the population impact of PWLs and standardized packaging. Although both tasks were shown independently, they were shown sequentially and potentially impacted by social desirability bias. Future work could randomize the order of the tasks and test the PWL messages in real-world settings. For the discrete choice task, we compared all experimentally manipulated packs versus a branded pack displaying with no PWL. This design provides external validity, but we did not compare each experimental pack to one another or with an option for no choice of either pack. Lastly, we limited the study to testing the effect of a single PWL image. Although we chose this image based on prior research,³⁰ future work should test other PWL variations and standardized packaging colors (e.g., green).

PWLs with gain- and loss-framed text were equally powerful at reducing appeal of cigarette packs in young adult cigarette smokers, especially when combined with standardized packaging. The findings add to evidence that PWLs and standardized packaging are important policy interventions to reduce smoking. The results can inform regulatory policy decisions for countries considering implementing PWL and standardized packaging regulations, or for those considering updating existing regulations.