Influence of Point-of-Sale Tobacco Displays and Graphic Health Warning Signs on Adults: Evidence From a Virtual Store Experimental Study

Abstract

Objectives. We tested the impact of banning tobacco displays and posting graphic health warning signs at the point of sale (POS).

Methods. We designed 3 variations of the tobacco product display (open, enclosed [not visible], enclosed with pro-tobacco ads) and 2 variations of the warning sign (present vs absent) with virtual store software. In December 2011 and January 2012, we randomized a national convenience sample of 1216 adult smokers and recent quitters to 1 of 6 store conditions and gave them a shopping task. We tested for the main effects of the enclosed display, the sign, and their interaction on urge to smoke and tobacco purchase attempts.

Results. The enclosed display significantly lowered current smokers’ (B = −7.05; 95% confidence interval [CI] = −13.20, −0.91; P < .05) and recent quitters’ (Β = −6.00, 95% CI = −11.00, −1.00; P < .01) urge to smoke and current smokers’ purchase attempts (adjusted odds ratio = 0.06; 95% CI = 0.03, 0.11; P < .01). The warning sign had no significant main effect on study outcomes or interaction with enclosed display.

Conclusions. These data show that POS tobacco displays influence purchase behavior. Banning them may reduce cues to smoke and unplanned tobacco purchases.

Smoking is the leading cause of preventable mortality in the United States, accounting for approximately 443 000 deaths each year.¹ Currently, 19.3% of US adults smoke cigarettes, and nearly half attempt to quit smoking each year.² However, approximately 90% of smokers who attempt to quit relapse within 6 months,^3–6 and relapses may occur years after quitting.⁷ The tobacco industry aggressively markets its products to consumers, spending nearly 90% of its $8 billion marketing budget on promotional allowances to retailers and advertising and price promotions at the point of sale (POS), making retail stores the most important advertising channel for the industry.⁸ Retail cigarette advertising and promotions have increased over time,^9,10 and cigarette products are prominently placed on shelves behind checkout counters, exposing all store customers to tobacco products, including youths and adults who do not smoke or have recently quit.

Tobacco ads and displays may act as cues to smoke,^11–13 stimulate purchases among customers who did not intend to buy cigarettes,^14,15 and influence relapse among recent quitters by stimulating cravings for cigarettes.¹⁶ In a telephone survey of Australian adults, Wakefield et al. found that 55.3% noticed POS displays often or always and 25.2% bought cigarettes as a result of seeing displays when shopping for something other than cigarettes.¹⁵ Among respondents who had attempted to quit smoking in the past 12 months, 37.7% reported that seeing the tobacco displays increased their urge to purchase cigarettes and 60.9% bought cigarettes even though they were trying to quit. Carter et al. conducted intercept surveys with shoppers outside supermarkets to examine the influence of tobacco displays at the time of purchase and found similar patterns: approximately 22% reported unplanned cigarette purchases, with nearly half (47%) influenced by tobacco displays.¹⁴ In a cohort study, Germain et al. found that smokers with moderate or high sensitivity to tobacco displays at baseline were significantly less likely to quit smoking at follow-up than were those with low sensitivity (moderate, odds ratio [OR] = 0.32; 95% confidence interval [CI] = 0.14, 0.74; P = .007; high, OR = 0.27; 95% CI = 0.08, 0.91; P = .035).¹⁶ Laboratory-based cue reactivity studies show that drug-dependent individuals react strongly to cues associated with past or current drug use, including nicotine; viewing images such as cigarette packs or other people smoking can elicit subjective cigarette craving and psychophysiological arousal (e.g., increased heart rate).^17,18

The federal Family Smoking Prevention and Tobacco Control Act of 2009¹⁹ gives state and local governments legal authority to regulate the time, place, and manner of tobacco advertising. To date, state and local governments have attempted to ban tobacco displays and mandate graphic health warning signs at the POS. Tobacco product displays have been banned in Ireland, Canada, and Australia, but not in the United States. Studies show that graphic antismoking advertising can elicit strong emotional responses from smokers and influence them to quit.²⁰ However, these studies have largely focused on media campaign advertising and cigarette pack warning labels, and it is unclear whether posting similar messages at the POS will have the same impact. In 2009, New York City required licensed tobacco retailers to post graphic warning signs at the POS. Coady et al. conducted street intercept surveys with adult smokers and recent quitters before and after policy implementation and found that signs increased awareness about the health risks of smoking and thoughts about quitting smoking but did not deter smokers from purchasing cigarettes.²¹ However, street intercept interviews are subject to social response bias and cannot adequately control for potential confounders.

To test the potential impact of these policies on US adults, we designed a virtual store experiment. Virtual reality applications simulate real-world environments and are useful for studying behavioral responses to environmental cues that may be difficult to assess in a real-life setting.^22–24 Virtual environments have been used to examine the impact of banning POS tobacco displays and ads on youths,²⁵ consumer food-purchasing decisions,²⁶ and the effects of smoking cues on cigarette cravings among adults.^27,28 Virtual reality studies examining cravings in smokers suggest that these techniques may be more effective and have stronger ecological validity than traditional methods (e.g., photos) for triggering and assessing craving.^28–32 We randomized adult smokers and recent quitters to virtual store conditions and had them conduct a shopping task to assess whether exposure to an enclosed tobacco product display and a graphic health warning sign decreased urges to smoke and tobacco purchase attempts.

METHODS

We designed a 3 × 2 experimental study with 3 variations of the tobacco product display (open, enclosed [not visible], enclosed with pro-tobacco ads) and 2 variations of a graphic health warning sign at the POS (present vs absent) for 6 total virtual store conditions (Appendix A, available as a supplement to the online version of this article at http://www.ajph.org). All conditions had tobacco advertisements posted throughout the store (e.g., checkout counter underhang, exterior of store door). We purchased an off-the-shelf model of a convenience store and extensively customized it with Unity 3D interactive gaming software (Unity Technologies, San Francisco, CA). We used similar methods in a related study with youths.²⁵

Participants

In December 2011 and January 2012, we recruited a national convenience sample of current smokers and recent quitters aged 18 years or older who already had or were willing to download the Unity 3D plug-in to run the virtual store application. We sent e-mail invitations to members of ResearchNow’s online adult panel and screened interested adults to determine eligibility.

We defined current smokers as persons who smoked cigarettes on some days or every day. We defined recent quitters as persons who did not currently smoke but had smoked regularly sometime in the past 12 months. Participants were blinded to the study conditions; we said that the study was about consumer shopping behavior, and we made no mention of tobacco displays or health warning signs.

Of 4189 eligible adults, 1216 completed the virtual shopping task and survey (Figure 1). We randomized approximately 200 participants to each of the 6 conditions (∼60% smokers and 40% recent quitters).

FIGURE 1—

Sampling framework of current smokers and recent quitters in virtual reality study of effects of point-of-sale (POS) tobacco displays and graphic health warnings: United States, December 2011–January 2012.

Virtual Retail Store Experiment

After they consented to participate, we instructed participants to select 4 items for purchase: a snack from the aisles, a drink from the coolers, and 2 items from the checkout counter area. At the checkout area, if the participant attempted to purchase cigarettes by clicking on the tobacco product display, the animated cashier asked, “Do you want to buy tobacco products?” If the participant clicked no, the cashier responded, “Okay.” If the participant clicked yes, the program displayed a list of brands for the participant to choose from. Once participants selected the final 2 items from the checkout counter, they were exposed to the display for another 10 seconds before the shopping task ended and they were directed to the survey.

The main study outcomes were tobacco purchase attempts in the virtual store and urge to smoke after completing the shopping task. We defined participants as attempting to purchase tobacco if they clicked on the tobacco product display and confirmed that they wanted to purchase tobacco. The Unity 3D software captured this information. The survey captured urge to smoke immediately after the shopping task by asking participants to type in a number in response to this item: “On a scale of 0 to 100, rate your urge to smoke after seeing the virtual environment with 0 being ‘no urge’ and 100 being ‘strongest urge I have ever experienced.’”²⁸

We also assessed whether participants saw tobacco products for sale in the virtual store, whether they saw the health warning signs, and the extent to which they thought the virtual store resembled a real convenience store (5-point Likert scale from not at all to a lot). To assess awareness of tobacco products in the virtual store, we asked participants, “Which of the following items did you see for sale in the virtual store?” Participants chose from a list of 5 actual products in the store and 2 distractors. We considered participants to have confirmed awareness of the sign if they responded affirmatively to the question, “When you were in the virtual store, did you notice health warning signs about smoking posted in the checkout area?” and identified more correct than incorrect messages from a list of 6 messages provided (Table 1).

TABLE 1—

Demographic Characteristics of Current Smokers and Recent Quitters and Virtual Shopping Task Measures in Virtual Reality Study of Effects of Point-of-Sale Tobacco Displays and Graphic Health Warnings: United States, December 2011–January 2012

Characteristic	Total (n = 1216), % or Mean (SD)	Condition 1: Open Display, No GHWS (n = 209), % or Mean (SD)	Condition 2: Enclosed Display, No GHWS (n = 200), % or Mean (SD)	Condition 3: Enclosed Display With Tobacco Ads on Cabinet, No GHWS (n = 210), % or Mean (SD)	Condition 4: Open Display, GHWS (n = 192), % or Mean (SD)	Condition 5: Enclosed Display, GHWS (n = 204), % or Mean (SD)	Condition 6: Enclosed Display With Tobacco Ads on Cabinet, GHWS (n = 201), % or Mean (SD)
Smoking status
Current smoker	61.1	61.7	61.0	61.9	58.9	62.8	60.2
Recent quitter	38.9	38.3	39.0	38.1	41.1	37.3	39.8
Age, y
18–34	37.4	36.8	35.0	34.0	42.7	37.9	38.0
35–54	35.6	36.8	36.0	35.4	32.8	31.5	41.0
≥ 55	27.0	26.3	29.0	30.6	24.5	30.5	21.0
Gender (male)	44.9	49.8	44.0	41.2	43.8	47.8	43.0
Race/ethnicity
White, non-Hispanic	87.1	88.8	85.9	90.9	86.2	84.3	86.4
Non-White, non-Hispanic	8.5	7.3	9.1	4.3	9.0	10.3	11.1
Hispanic	4.4	3.9	5.0	4.8	4.7	5.4	2.5
Education
≤ high school	10.8	8.1	15.0	10.5	7.8	14.2	9.0
≥ some college	89.2	91.9	85.0	89.5	92.2	85.8	91.0
Virtual shopping task
Time spent in shopping aisles, s	217.3 (3.5)	221.7 (7.9)	228.7 (10.7)	217.1 (8.6)	211.5 (7.3)	217.1 (9.4)	207.2 (7.6)
Time spent at counter, s	35.6 (0.6)	36.9 (1.5)	33.2 (1.5)	35.7 (1.5)	36.5 (1.5)	34.2 (1.3)	37.1 (2.0)
Virtual store resembled real-world convenience stores
Not at all	2.2	1.0	2.0	3.3	1.0	2.0	4.0
A little/somewhat	45.3	45.0	46.0	52.4	39.6	47.1	41.3
Very/a lot	52.5	54.1	52.0	44.3	59.4	51.0	54.7
Visit convenience stores few times/wk or practically every daya	44.6	40.4	44.0	46.2	42.7	44.1	50.3
Saw tobacco for sale	61.5	88.5	49.5b	47.6c	91.2	41.7d	51.7ef
Saw GHWSg	23.1	5.8	2.5	4.3	37.0h	52.9ij	37.8kl

Note. GHWS = graphic health warning sign.

^aDichotomized variable: Reference category is those who go to convenience stores about once per week, about once per month, or hardly ever.

^bP < .01 between conditions 1 and 2.

^cP < .01 between conditions 1 and 3.

^dP < .01 between conditions 4 and 5.

^eP < .01 between conditions 4 and 6.

^fP < .05 between conditions 5 and 6.

^gAwareness of GHWS confirmed if participants responded yes to "When you were in the virtual store, did you notice health warning signs about smoking posted in the checkout area?" and then correctly identified messages on the GHWS (Smoking causes lung cancer; Quit smoking today; Call 1-800-QUIT-NOW) from a list of 3 correct and 3 incorrect messages. If they identified only incorrect messages (Smoking causes impotence; Smoking causes birth defects; Call 311 to quit smoking) or identified more incorrect messages than correct messages, they were coded as not having confirmed awareness of the GHWS.

^hP < .01 between conditions 1 and 4.

ⁱP < .01 between conditions 2 and 5.

^jP < .01 between conditions 4 and 5.

^kP < .01 between conditions 3 and 6.

^lP < .01 between conditions 5 and 6.

Analysis

We linked data from the shopping task to the survey data via a unique identifier for each participant. We ran descriptive statistics and tested for potential differences by condition. We used regression methods to test for main effects of enclosed display and warning sign on urge to smoke (linear regression) and purchase attempt (logistic regression) outcomes. We also tested for an interaction to assess whether the effect of the enclosed display varied depending on whether the warning signs were present. To test the main and interactive effects of the enclosed display, we created a variable, enclosed display that combined all of the enclosed-display conditions (conditions 2, 3, 5, and 6).

In general form, we estimated the following regression equation:

where GHWS indicates graphic health warning sign.

We also ran additional models with a separate indicator for the ads on cabinet conditions, but this led to unstable estimates because this indicator was collinear with the enclosed-display variable.

We ran separate models for current smokers and recent quitters. All models controlled for potential confounders: age group, gender (male), race/ethnicity, education, and frequency of convenience store visits. Smoker models also controlled for time since last cigarette, time from waking until first cigarette, quit attempt in the past 12 months (yes or no), and plans to quit in the next 30 days (yes or no). Recent quitter models also controlled for time since last quit. We estimated descriptive statistics and regression models with Stata version 11.0 (StataCorp LP, College Station, TX).

RESULTS

Table 1 summarizes sample characteristics. The majority of study participants were aged 18 to 54 years, female, and non-Hispanic White, and had at least some college education. Demographic characteristics did not differ significantly across conditions.

Shopping Task, Perceptions, and Awareness

Virtual shopping task–related measures are summarized in Table 1. On average, participants took 217.3 seconds (3.6 minutes) to complete the shopping task, with about 35.6 seconds spent at the checkout counter. Most participants (52.4%) agreed that the virtual store resembled a real convenience store very much or a lot. More than 40% of participants visited convenience stores at least a few times per week.

Approximately 61.5% of participants reported seeing tobacco products for sale. Significantly more participants in the open-display conditions saw tobacco products for sale than in the enclosed conditions, regardless of whether tobacco ads were present on the enclosed cabinets (e.g., 88.5% for condition 1 vs 47.6% for condition 3; P < .01) or not (e.g., 88.5% for condition 1 vs 49.5% for condition 2; P < .01). As expected, confirmed awareness of the graphic health warning sign was higher when the sign was present (37.0%–52.9% for conditions 4–6) than when the sign was absent (2.5%–5.8% for conditions 1–3).

Urge to Smoke

Figure 2a summarizes urge-to-smoke outcomes. When a health warning sign was not present, participants in the open-display condition reported an average urge to smoke of 23.7 (SE = 2.1). The average urge to smoke was significantly lower in the enclosed-display condition (condition 2, mean = 17.7; SE = 1.9; P < .05) and in the enclosed-display condition that had tobacco ads on the cabinet (condition 3, mean = 16.9; SE = 1.8; P < .05). When a sign was present, the average urge to smoke did not differ significantly between the open-display condition (condition 4, mean = 22.8; SE = 1.9), and the enclosed-display condition (condition 5, mean = 21.7; SE = 2.0). However, urge to smoke was significantly lower in the enclosed-display condition that had tobacco ads on the cabinet (condition 6, mean = 15.3; SE = 1.7), than in the open-display condition (condition 6 vs 4, P < .01) and the enclosed-display condition (condition 6 vs 5, P < .05).

FIGURE 2—

Association of point-of-sale tobacco displays and graphic health warning signs (GHWSs) among current smokers and recent quitters with (a) urge to smoke and (b) purchase attempts: United States, December 2011–January 2012.

Note. For the urge to smoke, there was a significant difference (P < .05) between conditions 1 and 2, 1 and 3, and 5 and 6, and a significant difference (P < .01) between conditions 4 and 6. For purchase attempts, there was a significant difference (P < .01) between conditions 1 and 2, 1 and 3, 2 and 3, 4 and 5, and 4 and 6. Whiskers indicate 95% confidence intervals. Numbers in parentheses are standard errors.

^aUrge to smoke was captured by asking participants: “On a scale of 0 to 100, rate your urge to smoke after seeing the virtual environment with 0 being ‘no urge’ and 100 being ‘strongest urge I have ever experienced.’”²⁸

In nearly all conditions, the average urge to smoke was higher among participants who saw tobacco products for sale in the virtual store than among those who did not, but these differences were mostly nonsignificant (Appendix B, available as a supplement to the online version of this article at http://www.ajph.org). We found no statistically significant differences in urge to smoke between participants who were and who were not aware of the warning sign (Appendix B).

Tobacco Purchase Attempt

Figure 2b summarizes tobacco purchase attempts. When a health warning sign was not present, 38.3% of participants in the open-display condition (condition 1) attempted to purchase tobacco. Significantly fewer participants attempted tobacco purchases in the enclosed-display condition (condition 2, 2.0%; P < .01) and in the enclosed-display condition with tobacco ads on the cabinet (condition 3, 8.6%; P < .01). When a sign was present, 32.8% of participants in the open-display condition (condition 4) attempted to purchase tobacco. Significantly fewer participants attempted tobacco purchases in the enclosed-display condition (condition 5, 2.9%; P < .01) and in the enclosed-display condition with tobacco ads on the cabinet (condition 6, 5.0%; P < .01).

In nearly all conditions, participants who saw tobacco products for sale in the virtual store were more likely than those who did not to attempt tobacco purchases, but these differences were significant in only some conditions (Appendix B). We observed no statistically significant differences in tobacco purchase attempts between participants who were aware and who were not aware of the health warning sign (Appendix B).

Main and Interactive Effects

Table 2 presents regression results testing the main and interactive effects of the enclosed display and health warning sign on urge to smoke and tobacco purchase attempt outcomes. The enclosed display had a significant main effect on lowering urges to smoke among current smokers (B = −7.05; 95% CI = −13.20, −0.91; P < .05) and recent quitters (B = −6.00; 95% CI = −11.00, −1.00; P < .01) and on lowering tobacco purchase attempts among current smokers (adjusted OR [AOR] = 0.06; 95% CI = 0.03, 0.11; P < .01).

TABLE 2—

Regression Model Testing Main and Interactive Effects of Virtual Reality Enclosed Display and Graphic Health Warning Sign on Urge to Smoke and Tobacco Purchase Attempts Among Current Smokers and Recent Quitters: United States, December 2011–January 2012

	Urge to Smoke		Tobacco Purchase Attempts
Covariate	Current Smokers (n = 712), B (95% CI; SE)	Recent Quitters (n = 465), B (95% CI; SE)	Current Smokers (n = 712), AOR (95% CI)	Recent Quitters (n = 146), AOR (95% CI)
Main effect: enclosed display	−7.05* (−13.20, −0.91; 3.13)	−6.00** (−11.00, −1.00; 2.54)	0.06** (0.03, 0.11)	...
Main effect: GHWS	−0.14 (–7.44, 7.17; 3.72)	−1.26 (–7.01, 4.49; 2.92)	0.70 (0.41, 1.22)	0.81 (0.20, 3.33)
Interaction: enclosed display × GHWS	0.66 (−8.23, 9.56; 4.53)	0.71 (−6.33, 7.76; 3.58)	1.01 (0.42, 2.41)	...

Note. AOR = adjusted odds ratio; CI = confidence interval; GHWS = graphic health warning sign. Enclosed display incorporated all enclosed-display conditions (C2, C3, C5, C6). All regression models were adjusted for age, race/ethnicity, gender, education, and frequency of going to a convenience store. Current smoker models also controlled for time since last cigarette, time from waking until first cigarette, quit attempt in past 12 months, and plans to quit in the next 30 days. Recent quitter models also controlled for time since last quit.

*P < .05; **P < .01.

For recent quitters, we were unable to estimate the main effect of the enclosed display on purchase attempts because so few recent quitters in any condition made purchase attempts. Warning signs had no significant main effect on urge to smoke or tobacco purchase attempts for either current smokers or recent quitters. We observed no significant interaction between the enclosed display and warning sign.

DISCUSSION

In a virtual store, adult smokers reported lower urges to smoke and were less likely to purchase tobacco when tobacco product displays were enclosed (not visible). Adult recent quitters also reported significantly lower urges to smoke when displays were enclosed. Very few recent quitters tried to purchase tobacco, so we could not estimate the effect of the enclosed display on purchase attempts in this group. Overall, these results for adult smokers and recent quitters support findings from previous studies that tobacco displays may act as cues to smoke,^11–13 stimulate cravings for cigarettes among recent quitters,¹⁶ and stimulate unplanned tobacco purchases among smokers.^14,15

In our virtual store experiment, the addition of 1 graphic health warning sign did not affect urge to smoke or tobacco purchases. Participants in the sign conditions reported being aware of the sign, but results showed no evidence that it reduced or strengthened the effect of a display. Similarly, Coady et al. found that exposure to the New York City POS health warning signs did not deter smokers from purchasing cigarettes or help recent quitters stay abstinent.²¹ In our experiment, we placed 1 large sign prominently at the center of the tobacco display to mirror New York City’s mandate of posting 1 large sign where tobacco products are displayed or 1 small sign at each cash register.²¹ However, neither study may have provided adequate exposure to warning signs at the POS. We confirmed awareness of the sign through 1 sign and 1 virtual store visit, in contrast with Coady et al., who assessed exposure from store visits over the past month, and therefore sign awareness was substantially lower in our study (37% for condition 4) than in theirs (66%). Future studies should examine whether the number and placement of graphic health warning signs, different images or messages, and repeat exposure make these signs effective at the POS. Future studies should also examine the influence of health warning signs in the absence of tobacco advertising in retail stores, because we were not able to tease this out in our experiment. It is possible that the presence of tobacco advertising throughout the store (in our virtual experiment and also in the study by Coady et al.²¹) may have diminished the influence of the signs.

If our results from the virtual store translate to the real world, then policymakers should prioritize enclosing tobacco product displays rather than mandating graphic health warning signs at the POS. Smokers support policies banning POS tobacco displays and advertising,^14,33 and those intending to quit are more likely to do so.³³ Countries such as Ireland, Canada, and Australia have successfully banned POS tobacco displays,³⁴ but no local or state government in the United States has done so yet. In 2012, the Village of Haverstraw, New York, enacted a law to prohibit stores from openly displaying tobacco products.³⁵ However, tobacco manufacturers and the New York Association of Convenience Stores filed a federal lawsuit claiming that the law violated their freedom of speech to communicate with consumers and that the Federal Cigarette Labeling and Advertising Act preempted state and local governments from regulating cigarette advertising and promotions. Haverstraw withdrew its law to avoid a costly court battle. The tobacco industry and its allies used similar arguments to challenge New York City’s graphic health warning signs at the POS.³⁶ These legal tactics have been used by the industry to subvert marketing regulations in the past³⁴ and will likely be used to challenge New York City’s current efforts to prohibit tobacco product displays.³⁷

Strengths and Limitations

We used a randomized controlled study design, which minimized threats to internal validity and allowed us to use a virtual store to compare the potential impact of policies relative to the status quo environment. We kept tobacco ads posted throughout the store in all conditions to disentangle the impact of the tobacco product display from tobacco ads. It could be argued that enclosing the display reduced the number of ads that were on the display shelves, but these shelving unit ads were minimal relative to the overall display area and the large tobacco ads posted throughout the store. Our virtual store allowed us to immerse participants in interactive environments that simulated potential policy regulations and observe their behaviors. An observational measure of adults clicking on the tobacco product display and selecting a tobacco product to purchase may be a more valid measure of purchase intention than surveys and store intercept interviews that are prone to recall and social desirability bias. Finally, we confirmed that the participants thought the virtual store was a realistic representation of convenience stores they frequented.

Limitations included possibly insufficient participant exposure to the conditions. We attempted to provide substantial exposure to the display conditions by having adults select multiple products at the checkout counter and having the animated retailer keep the participant waiting by talking on his cell phone; however, the participants spent an average of 35.6 seconds at the checkout counter, which is considerably less than the 63 seconds adults typically spend at convenience store checkout counters waiting in line and paying for their purchases.³⁸ In addition, a 1-time exposure to a virtual store likely underestimates the cumulative impact of policy changes in the retail environment.

Participants may have been primed that the virtual task was about smoking and tobacco because the screening question asked about smoking status and the consent form mentioned that the survey would assess perceptions about smoking. However, the shopping task did not direct participants to purchase tobacco products, and, if priming occurred, it should have affected participants across all conditions because of randomization.

Our measure of urge to smoke was self-reported. We purposely asked participants about their urge to smoke as the first question in the survey after the shopping task was completed to immediately capture the influence of the display. Although this measure has been used in virtual reality studies,²⁸ we could not determine whether some other factor influenced participants’ urge to smoke, such as the presence of another smoker or tobacco products when participants were taking the survey. Finally, our results have limited generalizability because the ResearchNow panel is a convenience sample. However, because the purpose of our study was to test the potential impact of a display ban and graphic health warning sign at the POS, our primary concern was to minimize threats to internal validity. Using an online panel like ResearchNow allowed us to access a large number of adults who could be screened systematically to identify current smokers and recent quitters and to randomize them to the virtual store conditions via the Web.

Conclusions

In a virtual store environment, enclosing the tobacco product display can reduce urges to smoke among adult current smokers and recent quitters and deter virtual tobacco purchase attempts among adult current smokers. A graphic health warning sign had no main effect and neither countered the effect of a tobacco display nor strengthened the impact of an enclosed display.

If virtual store experiments translate to real-world experiences, then our results suggest that policymakers should prioritize enclosing tobacco displays rather than mandating signs at the POS. However, policymakers should consider potential legal challenges^39,40 and the potential unintended consequences (e.g., retailers posting pro-tobacco ads on top of enclosed cabinets) that may attenuate the intended policy outcomes.