Understanding whether price tag messaging can amplify the benefits of taxes: An online experiment

Abstract

Introduction:

Excise taxes on unhealthy products like sugary drinks and tobacco can reduce purchases of these products. However, little research has investigated whether messages at the point of purchase, such as enhanced price tags, can increase taxes’ effects by heightening psychological reactions. This study aimed to examine whether including messages about taxes on price tags could amplify the benefits of excise taxes on unhealthy products.

Methods:

In 2022, an online study recruited 1,013 US parents to view seven price tag messages (e.g., “includes a 19% sugary drink tax”) and a control (i.e., standard price tag with the tax included in the price) displayed in random order alongside sugary drinks. Participants were randomly assigned to view a caution-symbol icon or no icon on price tags. Analyses were conducted in 2023.

Results:

All seven messages discouraged parents from buying sugary drinks for their children compared to control (average differential effects [ADEs] ranged from .28–.48, all p<.001). All messages led to greater attention to the price tag (ADEs ranged from .24–.41, all p<.001) and greater consideration of the cost of sugary drinks (ADEs ranged from .31–.50, all p<.001). Icons elicited higher cost consideration than text-only price tags (ADE=.15, p<.010), but not discouragement (p=.061) or attention (p=.079).

Conclusions:

Messaging on price tags could make excise taxes more effective. Policymakers should consider requiring messaging on price tags when implementing taxes.

Introduction

Research demonstrates that taxation of unhealthy products is a powerful tool for promoting healthier behaviors.¹ On average, a 10% price increase is associated with a 4–5% reduction in tobacco use.² Similarly, meta-analyses estimate a 10% reduction in dietary intake of sugary drinks after a 10% price increase,³ and a 15% reduction in sales after tax implementation.⁴ Another meta-analysis found a reduction in demand for sugary drinks of 20% following the implementation of sugary drink taxes in US jurisdictions.⁵

Communication approaches at the point of purchase can change consumer behavior, including food purchasing.^6–11 One opportunity to leverage the impact of point-of-purchase communication would be to use the price tag itself to draw consumers’ attention toward the tax and further deter purchasing of sugary drinks. For example, in Mexico, only 46% of the public was aware of sugary drink taxes three years after implementation;¹² shelf labeling requirements could increase awareness of the existence of sugary drink taxes.

Enhanced price tags (i.e., price tags with messaging designed to increase attention to and processing of taxes at the point of display) could maximize the impact of taxes by drawing attention to the tax and subsequently eliciting larger reductions in purchases. Multiple studies have found that signaling price increases on price tags or menus appears to amplify the impact of the increases.^13–16 However, studies have not tested a wide range of price tag labeling options, such as displaying the percentage of the price comprised by the tax. Another unanswered question is whether including an icon on the price tag could amplify the benefits of sugary drink taxes in reducing purchases. Research suggests that icons can make warning labels more effective,^{11, 17–20} but this question has not been studied in the context of price tags.

This study aimed to examine whether enhanced price tags highlighting taxes discouraged sugary drink purchases, compared to tax-inclusive price tags with no messaging (i.e., standard price tags). The study also aimed to examine whether enhanced price tags elicited greater attention and made participants think more about the costs of sugary drinks, compared to standard price tags. Finally, the study examined whether including an icon on the price tags heightened reactions.

Methods

Study Sample

A sample of US adults in December 2022 was recruited using Qualtrics Market Research Survey Panel. Participants were eligible if they were 18 years or older, resided in the US, and were a parent or guardian of any children aged 2 to 12 years. Quotas were established to recruit at least 25% self-identifying as Hispanic, Latino, or Spanish and at least 25% self-identifying as Black or African American. Qualtrics’ quality control measures involved removing and replacing respondents who completed the survey in an implausibly short time, defined as completing in less than 1/2 of the median completion time. Additionally, the survey asked participants to report the day of the month of their child’s birthday twice, once at the beginning of the survey and once at the end of the survey. Qualtrics removed and replaced respondents who did not report matching days of the month for their child’s birthday. Finally, Qualtrics removed and replaced respondents who did not complete the entire survey. The total sample size after quality control measures was 1,013.

The study stimuli appear in Figure 1, Panel A. All price tags displayed the same price of $1.99. This price was decided by imposing a $0.02 per ounce tax, mirroring the highest in the US,²¹ on a 16-ounce beverage priced at $1.67, in line with common sugary drinks in the US based on Nielsen Homescan data. Price tags used eight different messages: 1) control (displaying the price only), 2) includes tax (“Includes a sugary drink tax”), 3) avoid extra cost (“Avoid the sugary drink tax by picking a different drink”), 4) save money (“Save money by picking a drink without added sugar”), 5) tax amount (“Includes a $0.32 sugary drink tax”), 6) tax percentage (“Includes a 19% sugary drink tax”), 7) nutrient warning (“WARNING: High in added sugar”), and 8) health warning (“WARNING: Sugary drinks increase risk of type II diabetes”).

Figure 1.

Stimuli used in study

Panel A. Price tag designs (between-subject factor) and price tag messages (within-subjects factor) tested in experiment

Panel B. Example of drinks with icon nutrient warning price tag displayed to participants. Branding is blurred for copyright reasons but was not blurred for participants.

Two versions of the eight price tags were created, one version with a yellow icon symbol (i.e., icon design) and one without the symbol (i.e., text-only design). The yellow icon was modeled after a 2019 proposal to require icon-plus-text health warnings on sugary drinks in California.²² All 16 price tags were displayed on an image of three sugary drinks on a shelf (Figure 1, Panel B). Three popular drink brands were selected to display a variety of sugary drink types.

The experiment used the Qualtrics survey platform. After screening for eligibility criteria, participants provided electronic informed consent. This experiment used a 2 × 8 between-within subjects experimental design, where price tag design (i.e., text-only or icon) was the between-subjects factor and price tag message (i.e., message displayed) was the within-subjects factor. The Qualtrics randomizer function randomized participants in a 1:1 allocation ratio to one of the two between-subjects factors (price tag design): text-only price tags or icon price tags (CONSORT diagram in Appendix Figure 1). Then, participants viewed the eight price tag messages displayed in random order within their randomly assigned arm and completed a survey. After completing the survey, participants received incentives in a reward type and amount set by the survey vendor (e.g., cash, reward points). The University of North Carolina Institutional Review Board approved the study procedures (#21-3135). Prior to starting data collection, the study design, measures, predictions, and analytic plan were registered on AsPredicted.org (https://aspredicted.org/X1L_PHL).

Measures

The primary outcome was discouragement from wanting to buy sugary drinks for one’s child (i.e., discouragement), assessed after each price tag. The measure, adapted from Baig et. al 2018,²³ read: “How much does this message discourage you from buying sugary drinks for your child?” Response options ranged from “not at all” (coded as 1) to “a great deal” (coded as 5). Discouragement, a single-item measure of perceived message effectiveness, was the primary outcome because perceived message effectiveness is often used to identify messages that have the potential to change behavior.²⁴ Perceived message effectiveness has been measured extensively in nutrition messaging research.^{17, 25–27} Perceived message effectiveness is generally sensitive enough to detect minor differences between similar messages, yet predictive of longer-term behavior change.^{23, 28}

The survey assessed two secondary outcomes after each price tag: attention to the price tag (“How much does this price tag grab you attention?”) and cost consideration (“How much does this price tag make you think about the cost of sugary drinks?”). Response scales for the secondary outcomes ranged from “not at all” (coded as 1) to “a great deal” (coded as 5). The secondary outcomes were selected because they tend to be predictive of behavior change in the context of messaging.^{29, 30} Then, the survey asked participants to select the price tag that most discouraged them from wanting to buy a sugary drink for their child, using a measure adapted from a prior study.³¹ Response options for this item were the eight price tags participants had previously viewed, displayed in random order. Finally, the survey assessed sociodemographic variables.

Statistical Analysis

Power calculations estimated that a sample size of ~1,000 participants was required to detect effects of f=.03 or larger between price tag messages. This small effect size is in line with prior message development studies that have found similarly-sized effects between messages.^{17, 26} These calculations assumed 80% power, sphericity, an alpha of .05, and a correlation between repeated measures of .50. Although the study was not powered to detect differences between price tag designs, the achieved effect size detectable was f=.07 or larger between price tag designs. Analyses were conducted in Stata MP version 17 in 2023, with two-tailed tests and a critical alpha of .05. Average differential effects (ADE; i.e., the average difference in predicted means between groups) are reported. Wald tests were used to examine the statistical significance of interaction terms.

Analysis of the primary outcome used mixed effects (multilevel) linear regression to account for the repeated measures at the participant level.³² These models regressed discouragement on indicators for the two factors tested (price tag message (Level 1 variable) and price tag design (Level 2 variable)) and their interaction. The models assumed the experimental factors to be fixed, while treating the intercept as random. Using the same approach, mixed effects models were run to examine the impact of experimental factors on attention and cost consideration.

Because none of the interaction terms were statistically significant (all p>.43), the final models included indicators for the two factors but not their interaction. Each price tag message type was compared to the control. Although the pre-registration specified pairwise comparison of intervention price tag messages, these analyses were not conducted to avoid a high number of statistical comparisons. Moderation analyses explored whether the impact of price tag design on discouragement differed by education and Latino ethnicity. These moderators were selected because prior studies have shown that the impact of taxes and front-of-package labels may vary by education and Latino ethnicity.^{17, 33, 34} For these analyses, separate mixed effects linear regression models regressed discouragement on price tag message, price tag design, the moderator, and the interaction between price tag design and the moderator.

Finally, the proportion of participants selecting each price tag as most discouraging was calculated. Although the pre-registered analytic plan stated that proportions would be compared within price tag design, findings were collapsed across designs because the pattern was very similar. The proportion of participants endorsing each tag type was compared to the next highest-endorsed tag, using a constant-only multinomial logit model.

Results

The total sample size was 1,013. The mean age of parents was 37 years and most were women (70%, Table 1). Just over one-third of the sample (36%) were white, 22% were Black or African American, and 17% were Hispanic or Latino. About two-thirds (64%) had less than a college degree. Approximately half of parents (46%) had an annual household income below $50,000 per year. Most parents (77%) consumed sugary drinks at least twice per week. Similarly, 72% of parents reported that their child between the ages of 2–12 had consumed sugary drinks at least twice per week.

Table 1.

Participant characteristics

Participant characteristics	All (n=1013)	Text-Only Price Tags (n=525)	Icon and Text Price Tags (n=488)
	n (%)	n (%)	n (%)
Age
18–25 years	92 (9%)	47 (9%)	45 (9%)
26–34 years	339 (33%)	163 (31%)	176 (36%)
35–44 years	392 (39%)	218 (42%)	174 (36%)
45–54 years	137 (14%)	67 (13%)	70 (14%)
Over 55 years	53 (5%)	30 (6%)	23 (5%)
Mean (SD)	37 (9)	37 (9)	37 (9)
Gender identity
Woman	708 (70%)	373 (71%)	335 (69%)
Man	302 (30%)	150 (29%)	152 (31%)
Non-binary	3 (0%)	2 (0%)	1 (0%)
Race/ethnicity
White	363 (36%)	190 (36%)	173 (35%)
Black or African American	226 (22%)	112 (21%)	114 (23%)
Hispanic or Latiro	176 (17%)	96 (18%)	80 (16%)
Asian	95 (9%)	46 (9%)	49 (10%)
American Indian or Alaska Native	32 (3%)	19 (4%)	13 (3%)
Middle Eastern or North African	1 (0%)	0 (0%)	1 (0%)
Native Hawaiian or other Pacific Islander	7 (1%)	2 (0%)	5 (1%)
Another race or ethnicity	0 (0%)	0 (0%)	0 (0%)
More than one race/ethnicity	112 (11%)	59 (11%)	53 (11%)
Education level
High school degree/GED or below	292 (29%)	152 (29%)	140 (29%)
Associate’s degree or some college/technical school	360 (36%)	196 (37%)	164 (34%)
Bachelor’s degree or higher	361 (36%)	177 (34%)	184 (38%)
Annual household income
$24,999 or less	220 (22%)	101 (19%)	119 (24%)
$25,000 to $49,999	247 (24%)	143 (27%)	104 (21%)
$50,000 to $74,999	202 (20%)	101 (19%)	101 (21%)
$75,000 to $99,999	153 (15%)	80 (15%)	73 (15%)
$100,000 or more	190 (19%)	100 (19%)	90 (18%)
# of people in household, mean (SD)	4 (1)	4 (1)	4 (1)
# of children under age18 years in household, mean (SD)	2 (1)	2 (1)	2 (1)
Frequency of SSB consumption
1 time/week or less	234 (23%)	114 (22%)	120 (25%)
2–3 times/week	326 (32%)	178 (34%)	148 (30%)
4–6 times/week	222 (22%)	118 (22%)	104 (21%)
1 or more times/day	231 (23%)	115 (22%)	116 (24%)
Age of child
2–5 years	376 (37%)	198 (38%)	178 (36%)
6–9 years	350 (35%)	180 (34%)	170 (35%)
10–12 years	287 (28%)	147 (28%)	140 (29%)
Mean (SD)	7 (3)	7 (3)	7 (3)
Gender identity of child
Girl	460 (45%)	231 (44%)	229 (47%)
Boy	553 (55%)	294 (56%)	259 (53%)
Another gender identity	0 (0%)	0 (0%)	0 (0%)
Race/ethnicity of child
White	350 (35%)	184 (35%)	166 (34%)
Black or African American	217 (21%)	108 (21%)	109 (22%)
Hispanic, Latino, or Spanish	169 (17%)	92 (18%)	77 (16%)
Asian	74 (7%)	36 (7%)	38 (8%)
American Indian or Alaska Native	27 (3%)	16 (3%)	11 (2%)
Middle Eastern or North African	0 (0%)	0 (0%)	0 (0%)
Native Hawaiian or other Pacific Islander	7 (1%)	2 (0%)	5 (1%)
Another race or ethnicity	0 (0%)	0 (0%)	0 (0%)
More than one race/ethnicity	168 (17%)	86 (16%)	82 (17%)
Frequency of SSB consumption of child
1 time/week or less	288 (28%)	142 (27%)	146 (30%)
2–3 times/week	372 (37%)	195 (37%)	177 (36%)
4–6 times/week	160 (16%)	84 (16%)	76 (16%)
1 or more times/day	193 (19%)	104 (20%)	89 (18%)

^aAsked about one child ages 2–12 with the most recent birthday.

Note. Participant we e allowed to select more than one race and ethnicity category as a response. Data were missing for 0.1% of values. Percentage totals may not equal 100% due to rounding.

As predicted, all enhanced tax price tags discouraged parents from wanting to buy sugary drinks for their children, compared to the tax-inclusive standard price tag control (all p<.001, Table 2; Appendix Table 1). Among the enhanced price tag messages, the “saving money” price tag had the smallest effect size (ADE=.28) and the “health warning” price tag had the largest effect size (ADE=.48). Similarly, all enhanced tax price tags led to greater attention to the price tags compared to the control (all p<.001, Table 2). The “includes tax” price tag had the smallest effect size for attention (ADE=.24) and the “health warning” price tag had the largest effect size (ADE=.41). Finally, all enhanced tax price tag messages led to greater cost consideration (all p<.001, Table 2). The “nutrient warning” price tag had the smallest effect size for cost consideration (ADE=.31) whereas the “tax percentage” price tag had the largest effect (ADE=.50).

Table 2.

Impact of enhanced sugary drink tax price tag messages and price tag designs (icon vs. text)

Experimental factor	Discouragement			Attention to price tags			Cost consideration
Price tag message	ADE	SE	p	ADE	SE	p	ADE	SE	p
Control (reference)	--	--	--	--	--	--	--	--	--
Includes tax	.33	.04	<.001	.24	.04	<.001	.41	.04	<.001
Avoid extra cost	.33	.04	<.001	.30	.04	<.001	.34	.04	<.001
Saving money	.28	.04	<.001	.27	.04	<.001	.37	.04	<.001
Tax amount	.38	.04	<.001	.32	.04	<.001	.42	.04	<.001
Tax percentage	.43	.04	<.001	.40	.04	<.001	.50	.04	<.001
Nutrient warning	.42	.04	<.001	.30	.04	<.001	.31	.04	<.001
Health warning	.48	.04	<.001	.41	.04	<.001	.41	.04	<.001
Price tag design
Text-only (reference)	--	--	--	--	--	--	--	--	--
Icon	.12	.07	.061	.11	.06	.079	.15	.06	.010

Note. ADE=average differential effect. SE=standard error for the ADE. Missing data ranged from .1 to .2%. Boldface indicates statistical significance (p<.05).

When selecting the price tag that most discouraged them from wanting to buy sugary drinks for their child (Figure 2), 37% of participants picked the health warning, followed by the nutrient warning (16%), tax percentage (14%), tax amount (13%), saving money (7%), avoiding extra costs (5%), control (4%), and includes tax (4%). Results from the multinomial logit model comparing each proportion to the next highest-endorsed price tag are depicted in Figure 2.

Figure 2.

Percentage selecting each price tag as the one that most discouraged purchasing sugary drinks (n=1009)

*Statistically significant results (p<.05) comparing the proportion endorsing each price tag to the next highest-endorsed tag (e.g. nutrient warning vs. health warning).

In terms of price tag design, icon price tags elicited greater cost consideration than text-only price tags (ADE=.15, p<.010). However, icon and text-only price tags did not differ on discouragement (ADE=.12, p=.061) or attention (ADE=.11, p=.079). Moderation analyses revealed that the impact of price tag design on discouragement did not differ by education (p=.707) or Latino ethnicity (p=.289).

Discussion

This study found that messaging on sugary drink price tags makes a difference in how excise taxes are perceived by potential purchasers. Enhanced price tags that conveyed information about a sugary drink tax outperformed a standard price tag that did not signal the presence of a tax. These enhanced price tags discouraged parents from wanting to buy sugary drinks for their child, drew more attention to the price tag, and made parents think more about the costs of sugary drinks. Adding a yellow caution symbol (i.e., icon) to the price tag led to greater thinking about the costs, but did not lead to greater discouragement or attention.

The finding that enhanced price tags outperformed standard price tags is in line with prior studies concluding that messaging can make consumers more responsive to price increases.^13–15 Taken together, these findings suggest that current tax policies are missing an opportunity by not considering formatting requirements for price tags at the point of sale. The current study builds on prior research by studying which kinds of enhanced price tag messages are likely to be the most effective. This study found that messages about added sugar content and health risks associated with sugary drinks were especially promising, in line with research.^{7, 34–40} Price tags including the amount of the tax ($0.32) and the percentage of the tax (19% price increase) performed similarly to each other, a surprising finding given that prior studies have reported challenges with interpretation of percentages when it comes to interpreting health risk information.^{41, 42} Messages about avoiding extra cost and saving money by not buying sugary drinks performed similarly to each other, but slightly worse than the messages including the amount and percentage of the tax.

This study suggests that policymakers have a range of enhanced price tag messages to consider. Currently, seven US cities and the Navajo Nation impose taxes on sugary drinks,²¹ leaving room for new policymaking in the rest of the US. In terms of the legal pathways for requiring price tag messaging, depending on the type of price tag, state and local governments could require price-related messaging on price tags via their authority to regulate economic activity without discriminating against interstate commerce,⁴³ or under their police power, which allows them to regulate in order to protect the public’s health and safety.⁴⁴ These price tag requirements could apply to retailers, while the tax itself is imposed on the distributor per current sugary drink tax regulation in the US. Price-related and health-related tags would be constitutional under the First Amendment as long as they are factual and noncontroversial, reasonably related to a legitimate government interest in promoting price transparency or improving consumer knowledge of health risks, and not unjustified or unduly burdensome to commercial speech.^45–48 This study found that warning-style messaging about health risks and added sugar content were among the highest-performing messages, suggesting the potential for product warning labels to amplify the benefits of taxes, even if warnings are implemented separately from tax policies. However, even the simplest message, “Includes a sugary drink tax,” outperformed standard tax-inclusive price tags on all outcomes in this study. Thus, policymakers have a range of potential approaches to consider. The bottom line is that calling out the presence of a tax on the price tag, rather than simply building the tax into the price, is likely to make consumers more sensitive to the tax. In response to new shelf tag requirements, it is possible that retailers would heighten marketing for sugary drinks or lower shelf prices to offset the taxes. Real-world studies could document industry responses to sugary drink taxation in general, including with additional shelf tag requirements.

Additionally, retailers might consider voluntary messaging about sugary drink taxes at the point of sale, with the goal of shifting consumers away from unhealthy products or perhaps inciting opposition to the tax. Indeed, one report found that 55% of stores surveyed in Seattle explicitly mentioned the sugary drink tax on the shelf tag.⁴⁹ An unanswered question is whether heightening attention to sugary drink taxes (either via voluntary signage or shelf tag requirements) would change public support for taxes. Future studies should examine the possibility that drawing attention to taxes could reduce public support, although it is also plausible that support could increase given prior research finding that public support often increases after implementation of a policy.⁵⁰

Including icons on price tags did not elicit differences in discouragement or attention, compared to text-only price tags. However, icons made people think more about the costs of sugary drinks. Prior studies have found that including icons in warnings makes them more effective than text alone.^{17, 18} But, these prior studies have tested health-focused icons,¹⁷ or icons in a within-subjects design¹⁸ which may have heightened differences in those studies.

Limitations

Strengths of this study include recruitment of a diverse sample of parents with respect to race, ethnicity, education, and income, as well as the use of real drink brands in the study stimuli to heighten realism. Limitations include the brief exposure to price tags, the within-subjects design that could have heightened differences between price tags, and the lack of a behavioral outcome; future studies should examine the impact of enhanced price tags on actual purchasing and consumption behavior in between-subjects experiments. Using a simplistic shelf with one price tag for three beverages could have heightened attention to the price tags. However, the effects in this study could be smaller than in the real-world since participants did not spend real money. Finally, this experimental study could not account for varying levels of tax awareness following implementation of a new tax.

Conclusions

Seven types of enhanced sugary drink price tags that called out the presence of the tax outperformed a standard price tag that simply included the tax in the price. Messaging about taxes on price tags could translate into larger reductions in purchases of unhealthy products, thus leading to greater health benefits. Policymakers should consider mandating formatting requirements for price tags to maximize the public health benefits of taxes to consumers.