Color and shape matter: how visual design features of front-of-package warning labels influence attractiveness perception, healthfulness perception and purchase intention

Abstract

Non-communicable diseases have become a global public health challenge, with excessive intake of sugar-sweetened beverages (SSBs) identified as a major dietary risk factor. To promote healthier consumption, front-of-package (FoP) nutrition labels has been widely implemented worldwide. Although China has not yet established a standardized FoP system, the Healthy China 2030 initiative demonstrates the government’s willingness to implement FoP labeling strategies as part of its effort to improve nutrition literacy and encourage healthier food choices. However, limited research has explored how the visual design of FoP warning labels (WLs) affects consumer perception and behavioral intention. To address this gap, the present study employed a 2 × 2 between-subjects quasi-experiment among Chinese Generation Z consumers, manipulating color (black and red) and shape (octagon and shield) of WLs on SSBs. Three dependent variables were measured: perceived attractiveness, perceived healthfulness, and purchase intention. Results showed significant main and interaction effects across all outcomes: red-octagon WLs were most visually attractive, black-octagon WLs most effectively conveyed unhealthfulness, and black-shield WLs most strongly discouraged purchase intention. These findings demonstrate that color and shape jointly shape the perceptual and behavioral impact of FoP WLs. Theoretically, based on cue utilization theory and food label information processing models, this study conceptually explains how visual warning cues influence consumers' perceptions and purchase intentions, and provides practical insights for developing evidence-based FoP policies to reduce SSBs consumption in China.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40519-026-01812-0.

Introduction

Non-communicable diseases (NCDs) have become one of the major global health problems. According to the World Health Organization (WHO), as of 2022, more than 2.5 billion adults worldwide were overweight, including 890 million who were obese, accounting for about 16% of the global adult population [1] The prevalence of obesity continues to rise rapidly, especially in developing countries, such as China [2, 3]. One key dietary factor contributing to obesity is the excessive intake of ultra-processed foods [4, 5]. Among ultra-processed foods, sugar-sweetened beverages (SSBs) are one of the largest sources of added sugars in the modern diet and are a major cause of various adverse health outcomes [6]. SSBs include any beverages containing added sugars, such as soft drinks, high-sugar fruit juices, and concentrated syrups. Frequent consumption of SSBs has been linked to weight gain, metabolic disorders, type 2 diabetes, and an increased risk of cardiovascular diseases [7, 8]. As a countermeasure, many countries have introduced front-of-package (FoP) labeling strategies over the past decade to help consumers make healthier choices and reduce the risk of NCDs [9–11]. FoP nutrition labels provide simplified nutritional information directly on food packaging and are recognized as an effective policy tool recommended by the WHO to guide healthier public dietary behaviors [12, 13], although their effectiveness in practice has been shown to vary across contexts [14].

The formats and regulatory approaches of FoP labels vary across countries. Voluntary labeling systems mainly include the NutrInform Battery, Guideline Daily Amounts, Traffic-light Labels, Nutri-Score, and Keyhole used in Europe, the Health Star Rating used in Australia, and the Facts up Front and Guiding Stars applied in the United States [15–19]. In contrast, mandatory labeling systems are mostly implemented in the form of warning labels (WLs). WLs highlight products with sugar or calorie levels exceeding a defined threshold and require the warning symbol to be displayed prominently on the front of the package [20]. Since Chile first introduced WLs in 2017, several countries in South America and North America have adopted similar policies [21]. Many evaluation studies have shown that WLs significantly improve consumers’ visibility and understanding of products high in sugar, and help discourage purchases of unhealthy foods and beverages, especially SSBs [22–25]. Empirical studies from Chile, Uruguay, Brazil, and Mexico have emphasized that WLs are highly attention-grabbing, easy to understand, and effective in communicating risks associated with high-sugar or high-fat foods [26–28]. Their findings indicate that the high salience and simplicity of warning symbols can reduce misperception and trigger consumers’ sense of risk. These characteristics have made WLs one of the most prominent FoP labeling systems worldwide.

Although China has one of the largest populations of overweight and obese individuals in the world, a standardized FoP nutrition labeling system has not yet been established [29, 30]. To address related public health challenges, the Chinese government has promoted healthy eating through the Healthy China Action and set a goal to reduce the intake of salt, saturated fat, and sugar by 2030. One key strategy is to encourage the food industry to provide supplementary nutrition information on the front of food packaging, to help consumers identify and choose healthier products [31, 32]. In the early stage of policy implementation, using SSBs as an initial product category and focusing on WLs, which have high visibility and are easy to understand, is more feasible in practice [26, 33, 34].

WLs usually consist of color, shape, and short warning text. By increasing information salience, they attract attention and trigger risk evaluation, which in turn influences consumers’ product perceptions and choices [26, 33]. Although previous studies have confirmed the overall effectiveness of WLs and other FoP labels [36–39], evidence on how specific visual design features shape consumer processing and responses remains limited, especially regarding the interaction effects between color and shape, which have not been directly tested empirically [40]. Existing research often examines color, shape, or text as single factors, with relatively few studies assessing the combined effects of key visual cues [26, 28, 41–43].

In addition, evidence on WLs visual cues is not fully consistent across countries and cultural contexts. For example, the octagon design, which is widely used in international practice, has been shown to be effective in some regions, while studies in the Chinese context suggest that the shield shape may be associated with stronger perceptions of authority and higher acceptance. Similarly, findings on the advantages of red and black colors in terms of perceived healthfulness and attention capture are mixed [26, 35, 43]. Therefore, it is necessary to further examine, within the Chinese consumer context, how the combined design of two core visual cues in WLs, color (black and red) and shape (octagon and shield), influences key consumer perceptions and behavioral intentions. This can provide targeted empirical evidence and design references for the upcoming implementation of FoP nutrition labeling policies in China.

The conceptual framework (Fig. 1) of this study is grounded in Cue Utilization Theory [44] and the Food Label Information Processing Model [45]. Cue Utilization Theory suggests that when product intrinsic cues, such as nutritional quality, cannot be directly observed, consumers rely on extrinsic cues, such as packaging or labels, to form evaluative judgments. In the context of WLs, visual design features such as color and shape act as salient external cues that convey health risk information and influence consumers’ initial product evaluations. Complementing this view, the Information Processing Model proposes that nutrition labels affect consumer decision-making through a series of stages, including exposure, perception, interpretation, and evaluation, which ultimately shape behavioral responses. From this perspective, highly salient visual cues are more likely to attract attention, facilitate rapid perceptual processing, and influence consumers’ understanding and evaluation of product healthfulness. Based on these theoretical perspectives, this study conceptualizes the color and shape of WLs as core visual cues influencing consumer perceptions, and treats perceived attractiveness, perceived healthfulness, and purchase intention as the dependent variables. Accordingly, this study adopted a 2 × 2 quasi-experimental design that systematically manipulated WL color (black and red) and shape (octagon and shield), while holding all other factors constant, to examine the main and interaction effects of these visual cues on perceptual and behavioral outcomes.

Fig. 1.

Conceptual model linking WL visual cues to consumer perceptions and purchase intention

Methods

Research design

This study employed a 2 × 2 between-subjects quasi-experimental design to examine the effects of WLs design elements, color (black and red) and shape (octagon and shield), on consumers’ perceived attractiveness of the labels, as well as their perceived healthfulness and purchase intention toward SSBs. Each participant was randomly assigned to one of the four experimental conditions, each corresponding to a specific combination of color and shape in the WLs.

Participants

A total of 272 participants took part in this study. All participants were urban young consumers residing in Beijing, China. This population was selected, because age-specific evidence indicates that individuals aged 13–29 in China consume significantly higher levels of SSBs and added sugars than older age groups [46], a pattern that coincides with the rapidly increasing prevalence of overweight and obesity among young people [2]. Moreover, Beijing reports one of the highest rates of overweight and obesity among Chinese adults, along with a significantly higher intake of SSBs compared with the national average [3, 47], making it a particularly relevant and representative context for examining FoP labeling effects and informing evidence-based policy implementation in China.

Participants were recruited through targeted online advertisements disseminated on major social media platforms in China. The advertisements invited interested individuals to contact the research team and voluntarily participate in the study. After participants indicated their willingness to take part in the study, all participants completed a short online background questionnaire for eligibility screening and subsequent descriptive statistical analysis. The questionnaire covered demographic information, socioeconomic status, self-reported body mass index (BMI), self-assessed nutrition knowledge, and attitudes toward healthy eating. Purchase frequency of SSBs, self-estimated nutrition knowledge, and attitudes toward healthy eating were all measured using Likert-type scales.

Eligibility criteria required participants to be urban residents of Beijing, aged between 18 and 30 years, and to report current consumption of SSBs. Exclusion criteria included: (1) formal education or professional experience in nutrition, health-related fields, or food design; (2) incomplete questionnaire responses; and (3) self-reported non-consumption of SSBs in the past 6 months. These criteria were applied to minimize potential biases arising from prior domain knowledge or limited consumption experience, thereby enhancing the internal validity of the study [48]. During recruitment, 28 participants were excluded due to relevant educational or professional backgrounds, and an additional 4 participants were excluded for reporting no SSBs consumption.

The required sample size for this quasi-experimental study was determined a priori using G*Power 3.1 [49]. Based on a 2 × 2 between-subjects design with fixed effects (main effects and interaction), the parameters were set to a medium effect size of f = 0.25 [50], a two-tailed of α = 0.05, and a power of 0.95. The analysis indicated a minimum total sample size of N = 210. The final analytical sample of 240 participants, therefore, exceeded this requirement, ensuring sufficient statistical power to detect both main and interaction effects.

In total, valid data from 240 participants were analyzed. After recruitment and screening, participants were randomly assigned to one of the four experimental conditions using a computer-generated randomization procedure, ensuring equal allocation across groups (n = 60 per group). All participants signed informed consent forms before the experiment. They were assured that their participation was voluntary and anonymous, and that there were no risks or conflicts of interest involved. All data were collected for academic research purposes only. This study received ethical approval in July 2025, with the approval code USM/JEPeM/PP/25020205.

Stimuli

The experimental stimuli included four WL designs (Fig. 2), which were created by crossing two visual design elements: color (black and red) and shape (octagon and shield). All WLs contained identical warning messages translated from Chile’s official “High in Sugar” label into Chinese. The size, text, font, and layout of all labels were kept fully consistent.

Fig. 2.

Four WLs stimuli created by crossing color and shape

The label designs and visual stimuli were reviewed by an expert committee through two rounds of evaluation. The first round involved review and revision, and the second confirmed the final designs. The committee consisted of six experts from the fields of public health, nutrition, and visual communication design. Based on their professional feedback, four label versions were finalized. The experts agreed that maintaining the key wording in Chinese would ensure direct communication of the message. The phrase “Ministry of Health” was added to the labels to enhance perceived authority and credibility.

All WLs were placed on the lower front section of the beverage packaging (Fig. 2) to ensure consistent visibility across all conditions. To minimize participants' biases toward familiar brands and packaging designs, SSBs were selected from regional beverage brands that were relatively unfamiliar to consumers in the Beijing area. Furthermore, brand logos and product names on the packaging were blurred to ensure that participants' evaluation responses stemmed from the visual appeal of the WLs, rather than their brand preferences [51, 52]. A pilot test involving 40 participants showed that virtualized SSB packaging was difficult to identify as a market product, effectively avoiding brand familiarity bias (Fig. 3).

Fig. 3.

Stimuli presentation: placement of WLs on the front of SSB packaging under four experimental conditions

Research instrument

To ensure environmental control and precision in data collection, the experiment was conducted in a sealed and soundproof room. The stimuli, consisting of front-view images of SSB products with different WL colors and shapes, were stored in separate computer folders by the research assistants to prevent participants from obtaining any prior information about the stimuli. All stimuli were displayed on identical 24-inch monitors with standardized brightness and color calibration to avoid regional color differences that might affect participants’ visual perception. Each participant viewed only the assigned stimulus individually under the same lighting conditions and at a fixed viewing distance to maintain consistency across sessions.

The three dependent variables in this study were measured using three validated questionnaires (see Table 1). All items were rated on a seven-point Likert scale (1 = strongly disagree, 7 = strongly agree). Minor adjustments were made to fit the SSB context, and all items were forward- and back-translated by bilingual experts to ensure accuracy and semantic equivalence. In the present study, the reliability analysis was re-conducted, and the results indicated good internal consistency for all scales.

Table 1.

Measurement items and reliability of the dependent variables

Variable	Items	Cronbach’s α	References
Perceived attractiveness	This label easily catches my attention	α = 0.907	[53]
	This label provides me with the information I need
	This label is easy to identify
	This label provides reliable information
Perceived healthfulness	I expect this product to be healthy	α = 0.944	[54]
	I would consider this product as good for me
	The product looks healthy
	This product looks low on Sugar
	I have an impression that this product is healthy
	This beverage looks healthier than similar products
Purchase intention	I would like to try this product	α = 0.886	[54]
	I would seriously consider buying this product
	I would buy this product

Data collection procedure

To avoid potential technical interference and data reliability issues that may occur in online experiments, such as network delay, repeated IP addresses, or automated bot responses [55], this study was conducted in a controlled offline environment. Before the experiment began, participants received detailed instructions about the research procedure to ensure full understanding and enhance the study’s credibility. Research assistants guided participants to their assigned seats and adjusted chair height to ensure consistent viewing distance and eye level with the screen.

Participants were randomly assigned to one of the four experimental conditions. Each participant viewed and evaluated only one stimulus to maintain independence in visual attention and perceptual evaluation, and to avoid potential fatigue or learning effects that could bias the results. Participants were allowed to freely view the assigned SSB packaging image for an average of 30 s. After viewing the stimuli, they completed a questionnaire evaluation. Once all tasks were finished, participants were debriefed and thanked for their participation. In quasi-experimental research, such a strictly controlled procedure ensures the precision and consistency of data collection [56].

Data analysis

The numerical data for perceived attractiveness, perceived healthfulness, and purchase intention were manually entered by two independent research assistants into Excel to prevent input errors and ensure accurate recording for subsequent statistical analysis. All data were analyzed using IBM SPSS 26.0. Descriptive statistics were used to summarize the quantitative results for the four experimental conditions, providing basic information of the data (including mean, standard deviation, skewness, and kurtosis) [57]. The internal consistency reliability was evaluated, and all Cronbach’s α values were above 0.7, indicating good reliability. The mean score of each variable was calculated by averaging the corresponding items (Table 1).

Prior to inferential analysis, the assumptions underlying analysis of variance were examined. Normality was assessed through inspection of residual distributions, and homogeneity of variance was evaluated using Levene’s tests. No substantial violations of these assumptions were detected.

To explore the main and interaction effects of the two design factors, two-way ANOVA was performed for each dependent variable. The main effect analysis examined the independent effects of each design factor (color and shape) on consumers’ perceptual and behavioral responses, clarifying how each factor separately influenced perceived attractiveness, perceived healthfulness, and purchase intention. The interaction effect analysis further examined whether the effect of one factor depended on the level of the other factor, revealing potential coordination between color and shape. When significant main or interaction effects were identified, post-hoc pairwise comparisons with Bonferroni adjustment were conducted to control for multiple comparisons and further examine group differences. In cases of significant interactions, simple effect analyses were conducted to test the influence of one independent variable at each level of the other variable, allowing a deeper understanding of the interaction patterns [58]. All statistical tests were performed using a two-tailed significance level of α = 0.05, and the effect sizes were reported using partial η² [50].

Results

A total of 240 valid samples were included in this study. Participants were randomly assigned to four experimental groups according to color (red and black) × shape (octagon and shield), with 60 participants in each group. Table 2 presents the detailed demographic characteristics of participants in each group. The chi-square test showed no significant differences among the groups in terms of all sociodemographic variables, SSBs purchasing frequency, and health-related variables (p > 0.05), indicating that random assignment was effective and that the groups were comparable [59].

Table 2.

Characteristics of the participants included in the randomized controlled trial

Characteristic	Red × Octagon (n = 60)	Black × Octagon (n = 60)	Red × Shield (n = 60)	Black × Shield (n = 60)	Total (n = 240)	Chi-square p value
Age, n (%)						0.632
18–22 y	22 (36.7)	18 (30.0)	20 (33.3)	25 (41.7)	72 (35.4)
23–26 y	21 (35.0)	21 (35.0)	25 (41.7)	16 (26.7)	83 (34.6)
27–30 y	17 (28.3)	21 (35.0)	15 (25.0)	19 (31.7)	72 (30.0)
Gender, n (%)						0.129
Male	26 (43.3)	35 (58.3)	33 (55.0)	24 (40.0)	118 (49.2)
Female	34(56.7)	25 (41.7)	27 (45.0)	36 (60.0)	122 (50.8)
Personal monthly income, n (%)						0.285
< ¥5162	19 (31.7)	24 (40.0)	24 (40.0)	27 (45.0)	94 (39.2)
¥5162–10248	23 (38.3)	20 (33.3)	28 (30.8)	20 (33.3)	91 (37.9)
> ¥10248	18 (30.0)	16 (29.1)	8 (13.3)	13 (23.6)	55 (22.9)
Education level, n (%)						0.705
Secondary school and lower	1 (1.7)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
High school	6 (10.0)	4 (6.7)	2 (3.3)	4 (6.7)	16 (6.7)
Undergraduate	47 (78.3)	50 (83.3)	54 (90.0)	49 (81.7)	200 (83.3)
Graduate	6 (10.0)	6 (10.0)	4 (6.7)	7 (11.7)	23 (9.6)
Purchase SSBs frequency, n (%)						0.148
Never	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Rarely	12 (20.0)	22 (36.7)	15 (25.0)	17 (28.3)	66 (27.5)
Sometimes	10 (16.7)	10 (16.7)	18 (30.0)	13 (21.7)	51 (21.3)
Often	17 (28.3)	19 (30.2)	14 (23.3)	13 (21.7)	63 (26.3)
Always	21 (35.0)	9 (15.0)	13 (21.7)	17 (28.3)	60 (25.0)
Self-reported BMI, n (%)						0.915
BMI < 18.5	6 (10.0)	7 (11.7)	7 (11.7)	10 (16.7)	30 (12.5)
18.5 ≤ BMI < 24	24 (40.0)	19(31.7)	21 (35.0)	17 (28.3)	81 (33.8)
24 ≤ BMI < 28	26 (43.3)	28 (30.8)	27 (45.0)	30 (50.0)	111 (46.2)
BMI ≥ 28	4 (6.7)	6 (10.0)	5 (8.3)	3 (2.4)	18 (7.5)
Previous diagnosis of chronic disease, n (%)
Diabetes	6 (10.0)	4 (6.7)	3 (5.0)	3 (5.0)	16 (6.7)	0.658
Hypertension	3 (5.0)	4 (6.7)	3 (5.0)	5 (8.3)	15 (6.3)	0.854
High cholesterol	8 (13.3)	8 (13.3)	6 (10.0)	6 (10.0)	28 (11.7)	0.886
High triglycerides	5 (8.3)	4 (6.7)	4 (6.7)	3 (5.0)	16 (6.7)	0.911
No above diseases	38 (63.3)	40 (66.7)	44 (73.3)	43 (71.7)	165 (68.8)	0.623
Self-estimated nutrition knowledge, n (%)a						0.966
No knowledge at all	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
Very little knowledge	3 (5.0)	3 (5.0)	4 (6.7)	3 (5.0)	13 (5.4)
Moderate knowledge	29 (48.3)	32 (53.3)	35 (58.3)	34 (56.7)	130 (54.2)
Considerable knowledge	16 (26.7)	13 (21.7)	13 (21.7)	15 (25.0)	57 (23.8)
Very thorough knowledge	12 (20.0)	12 (20.0)	8 (13.3)	8 (13.3)	40 (16.7)
Healthy eating attitude, n (%)b						0.186
Very bad	21 (35.0)	23 (38.3)	23 (38.3)	26 (43.3)	93 (38.8)
Not so good	8 (13.3)	8 (13.3)	7 (11.7)	7 (11.7)	30 (12.5)
Normal	11 (18.3)	8 (13.3)	10 (16.7)	10 (16.7)	40 (16.7)
Good	20 (33.3)	17 (28.3)	11 (18.3)	14 (23.3)	62 (25.8)
Very good	0 (0.0)	4 (6.7)	9 (15.0)	2 (3.3)	15 (6.3)

Bold values indicate Chi-square p values greater than 0.05, suggesting no statistically significant differences in participant characteristics across the experimental groups

^aData collected with the question: “In your view, how knowledgeable are you in nutrition?”

^bData collected with the question: “Which of the following statements best describes you?” with five response options adapted from [60]: (1) I like to eat whatever I enjoy and do not worry about whether it is healthy; (2) I try to choose healthy foods but find it hard to keep the habit; (3) I am neutral about healthy eating; (4) I consider taste/quality first and health second; (5) I eat healthy foods, because they help me stay healthy

Perceived attractiveness

Results from the two-way ANOVA (see Table 3) indicated that both color (p < 0.001) and shape (p < 0.05) had significant main effects on perceived attractiveness, as well as a significant interaction between the two factors (p < 0.001). In terms of color, participants exposed to red WLs reported higher perceived attractiveness (M = 5.59, SD = 1.38) than those exposed to black WLs (M = 4.40, SD = 1.64). With respect to shape, octagon WLs were rated as more attractive (M = 5.20, SD = 1.62) than shield WLs (M = 4.79, SD = 1.61). The significant interaction further indicated that the effect of color on perceived attractiveness varied across label shapes.

Table 3.

Between-subjects effects for perceived attractiveness

Source	df	F	p	Partial η2
Color	1, 236	40.21	< 0.001	0.146
Shape	1, 236	4.66	0.032	0.019
Color × shape	1, 236	15.09	< 0.001	0.06

Bold p values indicate statistically significant between-subjects effects (p ‹ 0.05)

To clarify this interaction (illustrated in Fig. 4), we conducted simple effect analyses using Bonferroni-adjusted comparisons (Table 4). For black WLs, no significant difference in attractiveness emerged between the octagon and shield shapes (p = 0.223). However, for red WLs, the octagon shape was rated significantly higher than the shield (p < 0.001). Furthermore, when comparing colors within each shape, red WLs were perceived as significantly more attractive than black WLs in the octagon condition (p < 0.001). In contrast, no significant color effect was detected for the shield condition (p = 0.084).

Fig. 4.

Interaction effect of perceived attractiveness

Table 4.

Simple effect analysis for the interaction between color and shape on perceived attractiveness

Variable	I	J	Mean difference (I − J)	F	p
Color
Black	Octagon	Shield	− 0.325	1.490	0.223
Red	Octagon	Shield	1.138	18.256	< 0.001
Shape
Octagon	Black	Red	− 1.925	52.282	< 0.001
Shield	Black	Red	− 0.463	3.018	0.084

Bold p values indicate statistically significant simple effects (p ‹ 0.05)

Perceived healthfulness

In Table 5, the two-way ANOVA revealed significant main effects of color and shape on perceived healthfulness (p < 0.001). The main effect of color showed that SSBs with red WLs were perceived as less healthy (M = 5.63, SD = 1.23) than those with black WLs (M = 3.75, SD = 1.65). The main effect of shape indicated that SSBs with shield WLs were perceived as healthier (M = 5.24, SD = 1.46) than those with octagon WLs (M = 4.13, SD = 1.81). In addition, the color × shape interaction was statistically significant (p < 0.05).

Table 5.

Between-subjects effects for perceived healthfulness

Source	df	F	p	Partial η2
Color	1, 236	118.28	< 0.001	0.334
Shape	1, 236	41.29	< 0.001	0.149
Color × shape	1, 236	4.7	0.031	0.02

Bold p values indicate statistically significant between-subjects effects (p ‹ 0.05)

The interaction pattern is illustrated in Fig. 5. To further examine this interaction, Bonferroni-adjusted simple effects analyses were conducted (Table 6). Under the black WLs condition, a significant difference was observed between shapes, with the octagon WLs receiving lower perceived healthfulness scores than the shield WLs (p < 0.001). A similar shape effect was also found under the red WLs condition, where the octagon WLs again received significantly lower perceived healthfulness scores than the shield WLs (p = 0.003). When examining the effect of color within each shape, red WLs were associated with significantly lower perceived healthfulness scores than black WLs under both the octagon (p < 0.001) and shield (p < 0.001) conditions.

Fig. 5.

Interaction effect of perceived healthfulness

Table 6.

Simple effect analysis for the interaction between color and shape on perceived healthfulness

Variable	I	J	Mean difference (I − J)	F	p
Color
Black	Octagon	Shield	− 1.486	36.933	< 0.001
Red	Octagon	Shield	− 0.736	9.061	0.003
Shape
Octagon	Black	Red	− 2.256	85.078	< 0.001
Shield	Black	Red	− 1.506	37.905	< 0.001

Bold p values indicate statistically significant simple effects (p ‹ 0.05)

Purchase intention

As shown in Fig. 6 and Table 7, the two-way ANOVA results indicated significant main effects of color (p < 0.001) and shape (p < 0.05) on purchase intention, as well as a significant interaction between color and shape(p < 0.001). The main effect of color showed that SSBs with black WLs produced lower purchase intention (M = 3.74, SD = 1.71) than those with red WLs (M = 5.64, SD = 1.23). The main effect of shape indicated that shield WLs were more effective in reducing purchase intention (M = 4.42, SD = 1.91) than octagon WLs (M = 4.96, SD = 1.57).

Fig. 6.

Interaction effect of purchase intention

Table 7.

Between-subjects effects for purchase intention

Source	df	F	p	Partial η2
Color	1, 236	113.77	< 0.001	0.325
Shape	1, 236	9.27	0.003	0.038
Color × shape	1, 236	33.21	< 0.001	0.123

Bold p values indicate statistically significant between-subjects effects (p ‹ 0.05)

Simple effect analyses (Table 8) showed that, under the black WL condition, shield labels elicited significantly lower purchase intention than octagon-shaped labels. Under the red WL condition, no significant difference in purchase intention was found between the two shapes. When the effect of color was examined within each shape, red WLs consistently resulted in higher purchase intention than black WLs.

Table 8.

Simple effect analysis for the interaction between color and shape on purchase intention

Variable	I	J	Mean difference (I − J)	F	p
Color
Black	Octagon	Shield	1.567	38.789	< 0.001
Red	Octagon	Shield	− 0.483	3.692	0.056
Shape
Octagon	Black	Red	− 0.872	12.023	< 0.001
Shield	Black	Red	− 2.922	134.952	< 0.001

Bold p values indicate statistically significant simple effects (p ‹ 0.05)

Discussion

This study adopted a 2 × 2 quasi-experimental design to examine how WL design elements, color (black and red) and shape (octagon and shield), influence young Chinese consumers’ perceptions and purchase intentions toward SSBs under controlled experimental conditions. The results showed when compared with black WLs, red WLs received higher scores for perceived attractiveness, especially the red-octagon label, which achieved the highest attractiveness rating. However, this higher attractiveness did not translate into lower perceived healthfulness or lower purchase intention when compared with black WL conditions, suggesting a potential divergence between visual appeal and warning effectiveness. In contrast, relative to red WLs, black WLs produced relatively low ratings in both perceived healthfulness and purchase intention across both shape conditions. Regarding shape, octagon WLs received slightly higher attractiveness scores than shield WLs when directly compared within the same color condition and were also perceived as healthier, whereas shield WLs were more effective than octagon WLs in reducing purchase intention under otherwise identical visual settings.

Effects of color on perceived attractiveness and healthfulness

The findings show that WLs color played a significant role in shaping young Chinese consumers’ perceptual responses to SSBs packaging. Specifically, red WLs were rated as more visually attractive than black WLs. However, compared with black WLs, red WLs were also associated with higher perceived healthfulness and higher purchase intention scores, suggesting that greater visual salience did not translate into stronger deterrent effects under the experimental conditions.

This pattern can be interpreted in light of the color-in-context theory proposed by Elliot and Maier [61], which posits that colors acquire meaning through both biological predispositions and culturally learned associations, thereby influencing attention and emotional processing. Red, as a highly salient signal color, has been shown to cause quick emotional arousal and promote early stage attentional capture [62]. In the context of WLs, this heightened arousal may cause red labels to appear more eye-catching and visually appealing compared with black labels. Consistent with this interpretation, previous studies have reported that red WLs attract more visual attention than darker ones [26, 43].

However, the present findings also highlight the dual psychological meaning of red. Prior research suggests that red can activate both avoidance-related responses (danger, threat) and approach-related responses (excitement, stimulation), depending on contextual cues [63, 64]. In avoidance-oriented contexts, such as hazard communication, red typically signals warning and risk. In contrast, in approach-oriented contexts, such as marketing and product packaging, red is often associated with attractiveness and stimulation. Because the SSB packages in this study simultaneously contained warning information and marketing elements, participants may have interpreted red WLs, particularly the red-octagon WL, as having a promotional or marketing-related meaning rather than functioning purely as warning cues. As a result, red WLs were perceived as more attractive, while their capacity to reduce perceived healthfulness remained weaker compared with black WLs.

Taken together, these findings suggest that, within the experimental context of this study, color alone does not uniformly determine warning effectiveness. While red WLs were more visually attractive than black WLs, black WLs were more strongly associated with lower perceived healthfulness. This contrast underscores the importance of considering the contextual and symbolic meanings of color when designing WLs, particularly in environments, where warning cues coexist with promotional packaging elements. Importantly, these interpretations are specific to young Chinese consumers and should be understood within the cultural and experimental context of the present study.

Effects of shape on perceived healthfulness

The results further demonstrate that WLs shape significantly influenced young Chinese consumers’ perceived healthfulness of SSBs. Specifically, octagon WLs were associated with lower perceived healthfulness ratings than shield WLs, indicating a stronger warning signal relative to the alternative shape. This effect was particularly pronounced when the octagon shape was combined with a black background, which produced the lowest perceived healthfulness scores among all experimental conditions.

These findings are consistent with earlier studies on warning symbol design, which suggests that angular and unstable shapes are more likely to be interpreted as signals of danger. For example, Riley et al. [65] identified the octagon as one of the most recognizable warning shapes due to its sharp edges and visual instability, characteristics that tend to evoke alertness and risk perception. Similarly, Cabrera et al. [26] reported that octagon WLs were more strongly associated with unhealthiness than alternative shapes, and that participants responded more rapidly to octagonal configurations, suggesting efficient processing of risk-related cues. In this study, the lower perceived healthfulness associated with octagon WLs relative to shield WLs aligns with this stream of evidence.

These results can be understood in light of Cue Utilization Theory [44]. When intrinsic product attributes such as nutritional quality are not directly observable, consumers rely on salient extrinsic cues to form evaluative judgments. Shape, as a highly accessible visual cue, appears to function as a heuristic signal of product risk. In this study, the octagon shape conveyed a stronger unhealthiness cue than the shield shape, consequently lowering perceived healthfulness under experimental setting.

In contrast, shield-shaped WLs were associated with comparatively higher perceived healthfulness ratings. This pattern may reflect the symbolic ambiguity of the shield shape, which is commonly associated with protection, security, or endorsement rather than prohibition. While such associations were not directly measured in the present study, prior qualitative evidence among Chinese consumers suggests that shield symbols can evoke perceptions of authority and trustworthiness [35]. Within the framework of the food label information processing model [45], this uncertainty may slow down the early stage risk appraisal process, leading to smaller drops in how healthy the product seems compared to the more menacing octagon shape.

Interaction effects on purchase intention

Beyond the independent effects of color and shape, the present study showed a significant interaction between these two visual design elements on purchase intention. Specifically, the black-shield WL configuration was associated with lower purchase intention than the other WL conditions, whereas red WLs, regardless of shape, were associated with comparatively higher purchase intention under experimental conditions. This interaction pattern suggests that the behavioral implications of WL color depend on how it is combined with specific shape cues, rather than operating independently.

Previous research on WLs has predominantly examined individual design features, such as color, shape, or text, as separate determinants of consumer responses [26, 28, 43]. In contrast, the present findings suggest that purchase intention is particularly sensitive to the joint configuration of visual cues, consistent with the information processing framework proposed by Wogalter and Laughery [66], which describes a progression from attention capture and risk communication to cognitive evaluation and subsequent behavioral intention. In this process, WLs first convey health risk and then activate consumers’ motivation to protect their health, thereby reducing purchase intention.

Within this framework, the black-shield WL configuration may have conveyed a more integrated and unambiguous warning signal than other combinations. Black backgrounds are commonly associated with seriousness, danger, and threat in hazard communication [67], whereas shield symbols are widely linked to protection, authority, and institutional endorsement [68]. When combined, these cues may have reinforced a consistent interpretation of health risk and self-protection, strengthening their deterrent effect on purchase intention. In contrast, red WLs, particularly when paired with shapes lacking strong prohibitive or authoritative connotations, may have generated more ambiguous interpretations, attenuating their effectiveness at the behavioral intention stage.

The observed interaction also underscores the role of cultural context in shaping cue interpretation. Although octagon WLs have demonstrated strong deterrent effects in several Latin American countries [21, 69], qualitative evidence suggests that Chinese consumers may attribute greater credibility and authority to shield-shaped warnings [35]. In this cultural setting, shield-shaped cues may, therefore, be more effective than octagonal cues in translating perceived risk into reduced purchase intention. This pattern aligns with the broader effect of cultural consistency, where warning designs that align with embedded connotations of authority and trust within a culture are more effective at evoking behavioral constraints [70, 71]. Nevertheless, this interpretation should be treated with caution, because this study did not directly measure cultural connotations and evaluation processes.

Implications for FoP WLs design in the Chinese context

The present findings offer several implications for the design of FoP WLs within the Chinese context. First, the observed differences across color and shape configurations suggest that visual salience alone is insufficient to influence behavioral intention. While red WLs were more visually attractive, they were not consistently associated with lower perceived healthfulness or reduced purchase intention. In contrast, black WLs, particularly when combined with shapes conveying clear warning or authority signals, were more strongly associated with lower purchase intention. From a design perspective, this indicates that FoP labels should prioritize visual clarity and risk communication over attention-grabbing aesthetics, especially for products with established marketing appeal, such as SSBs.

Second, the interaction effects between color and shape underscore the need to treat WLs as integrated visual systems rather than as collections of independent elements. The findings suggest that certain combinations of visual cues may facilitate more coherent interpretations of health risk, whereas others may generate mixed or ambiguous signals. For policymakers and designers, this implies that FoP label guidelines should specify not only individual design components but also recommended combinations of color and shape that align with the intended warning function.

Importantly, these implications should be interpreted within the specific cultural and regulatory context of China. Evidence from other regions, particularly Latin America, demonstrates that black-octagon WLs can be effective deterrents in real-world settings. However, qualitative research among Chinese consumers suggests that alternative symbols, such as shield shapes, may convey stronger perceptions of authority and credibility [35] The experimental evidence provided by this study suggests that black and shield shapes may play an important role in influencing the purchasing intentions of young Chinese consumers, but further verification in real-world environments is still needed.

Limitations and future recommendations

Despite its valuable contributions, this study acknowledges several limitations, which also suggest directions for future research. First, as a controlled visual stimulus experiment, participants’ behavioral responses were measured by self-reported purchase intention rather than actual purchasing behavior. To address this limitation, future studies could simulate realistic retail environments through virtual supermarkets to measure actual purchase behavior and enhance external validity. Second, this study employed a quasi-experimental design without a no-label control condition. As a result, the present findings do not allow for quantifying the absolute effectiveness of WLs in steering consumer choices compared with a baseline condition without FoP labels. Instead, the results should be understood as showing relative differences between different color and shape designs within the WLs category. To evaluate the overall effectiveness of WLs on consumer behavior, future studies should include a no-label control group as a baseline for comparison.

Third, the sample of this study was limited to urban Generation Z consumers in Beijing. While this group represents a high-risk population for SSB consumption, future research could include broader age groups and conduct cross-regional comparisons to further examine the generalizability of the findings across different demographic and cultural contexts in China. Fourth, by integrating eye-tracking technology, physiological measures, and qualitative methods, future research could more comprehensively explore how FoP labels influence attention, emotion, and decision-making processes across multiple stages of information processing.

Finally, this study focused specifically on WLs design elements and did not directly compare WLs with other FoP labeling systems, such as NutrInform Battery, Traffic-Light Labels, or the Healthier Choice Symbol. These systems employ distinct visual cues, such as graded indicators, color coding, or graded health symbols [72, 73], which may trigger different cognitive and emotional responses. Future randomized controlled trials or virtual supermarket experiments in the Chinese context are, therefore, needed to systematically compare the effectiveness of different FoP labeling formats and their downstream implications for dietary choices and public health outcomes.

Conclusion

This study employed a 2 × 2 quasi-experimental design to investigate how key visual design elements of FoP WLs, specifically color and shape, influence young Chinese consumers’ perceptions and purchase intentions toward SSBs under controlled experimental conditions.

Practically, the findings provide concrete guidance for policymakers and regulatory agencies involved in the development of FoP WLs standards in China. The findings indicate that policy recommendations should prescribe integrated combinations of visual elements, rather than addressing individual design components in isolation, to effectively communicate health risk. In particular, black backgrounds combined with shapes conveying clear warning or authority signals (octagon or shield) were more consistently associated with lower perceived healthfulness and reduced purchase intention among young Chinese consumers.

From a policy perspective, these findings support the inclusion of evidence-based visual design specifications, such as recommended color and shape associations, within future FoP labeling regulations, especially for high-risk product categories like SSBs. Such specifications could be piloted in regulatory sandbox programs or regional trials before nationwide implementation. In this way, the present study offers empirically grounded design parameters that may inform the gradual development of FoP labeling policies under China’s “Healthy China 2030” while acknowledging the need for further validation in real-world retail environments.

Theoretically, this study contributes by clarifying and operationalizing how established theories can be jointly applied to understand consumers’ responses to WLs. By integrating Cue Utilization Theory [44] with the Food Label Information Processing Model [45], the study conceptually distinguishes the role of visual design features as salient extrinsic cues that operate primarily at the early perceptual and evaluative stages of information processing. Specifically, color and shape are shown to function as heuristic signals that shape perceived attractiveness and perceived healthfulness, which in turn are associated with purchase intention under controlled experimental conditions.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. Methodology was developed by Zhiyi Guo and Yueyue Ning, and validation was conducted by Zhiyi Guo, Yueyue Ning, and Muhizam Mustafa. Formal analysis, investigation, data curation, and visualization were performed by Zhiyi Guo. The first draft of the manuscript was written by Zhiyi Guo, and all authors contributed to review and editing. Supervision was provided by Muhizam Mustafa. Project administration and funding acquisition were carried out by Zhiyi Guo. All authors read and approved the final version of the manuscript.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The data sets generated and analyzed during the current study are not publicly available due to ethical and confidentiality restrictions but are available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethical approval

All procedures performed in studies involving human participants were conducted in accordance with the ethical standards of the institutional and national research committee, the Declaration of Helsinki (2013 revision), and the Administrative Measures for Ethical Review of Life Science and Medical Research Involving Human Beings issued by the State Council of the People’s Republic of China (2023). Ethical approval for this study was obtained from the Human Research Ethics Committee of Universiti Sains Malaysia (USM/JEPeM/PP/25020205).

Inform consent

All participants were informed about the purpose and procedures of the study and provided written informed consent prior to participation. Participation in the study was entirely voluntary, and participants were free to withdraw at any time without penalty or loss of benefits.