Reformulation pathways shape the impact of front-of-pack nutrient grades: A discrete choice experiment on sweetener substitution

Reformulation pathways shape the impact of front-of-pack nutrient grades: A discrete choice experiment on sweetener substitution | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Reformulation pathways shape the impact of front-of-pack nutrient grades: A discrete choice experiment on sweetener substitution Yirui Chen, Hongxin Gui, Zhiyuan Liu, Kai Ma, Tieniu Zhao, Mengyang Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8573773/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Front-of-pack (FOP) nutrient grading systems are intended to shift demand toward healthier products and to incentivize reformulation. In beverages, however, favorable grades are often achieved through sugar reduction via artificial sweetener substitution, potentially creating a credibility conflict between an algorithmic "health" signal and ingredient cues that consumers interpret as artificial or non–clean label. This study tests whether the demand-side effectiveness of nutrient grades is conditional on reformulation pathways. We conducted a discrete choice experiment with 2,736 Chinese adults (12 choice tasks; opt-out included) in which generic 330 mL carbonated soft drinks varied by Nutri-Grade, sweetener strategy, flavor, and price. Random parameters logit models show a robust grade–ingredient interaction consistent with conditional signaling: the Grade A premium is substantially attenuated when paired with the artificial sweetener blend, flattening predicted choice gains from grade improvements. In monetary terms, willingness-to-pay for upgrading from Grade C to Grade A is + 2.31 RMB under cane sugar but becomes statistically indistinguishable from zero (+ 0.27 RMB) under artificial sweeteners, implying a valuation erosion of about 2.04 RMB. Latent class models reveal strong heterogeneity: a large "clean-label purist" segment (44.2%) drives grade discounting under artificial sweeteners, whereas "algorithmic health seekers" (32.5%) respond strongly to grades with minimal discounting. The findings indicate that nutrient grades function as conditional policy signals whose effectiveness depends on the reformulation strategies used to obtain favorable scores, with implications for how FOP grading policies are designed and evaluated in categories where additive-salient reformulation is prevalent. Physical sciences/Mathematics and computing Biological sciences/Psychology Social science/Psychology Front-of-pack labeling Nutrient profiling Reformulation incentives Artificial sweeteners Clean label Signal credibility Discrete choice experiment Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction 1.1. Front-of-pack nutrient grading as a policy instrument Excess intake of free sugars is widely recognized as a major modifiable risk factor for obesity, type 2 diabetes, and other non-communicable diseases. Sugar-sweetened beverages are among the most prominent contributors to added sugar intake and, as a result, have become a central target of nutrition-related public policy interventions. In recent years, front-of-pack labeling (FOPL) systems—particularly nutrient grading or scoring schemes—have been increasingly adopted or proposed as policy tools to improve dietary choices and to incentivize product reformulation. Nutrient grading systems aim to simplify complex nutritional information into salient, easy-to-interpret signals that can guide consumer choice at the point of purchase. By translating nutrient profiles into summary grades or scores, these systems are expected to operate through two complementary channels: first, by shifting consumer demand toward products with more favorable grades; and second, by encouraging manufacturers to reformulate products in order to achieve higher grades [ 1 – 3 ] . From a policy perspective, the effectiveness of nutrient grading thus depends not only on consumers' responsiveness to the label itself, but also on how firms strategically adapt their product formulations in response to the scoring criteria. Empirical evidence suggests that front-of-pack nutrient labels can influence purchasing behavior and improve the nutritional quality of food environments. However, growing research also indicates that the impact of these labels is heterogeneous across products, consumer groups, and implementation contexts [ 4 – 5 ] . This raises a critical policy question: under what conditions do nutrient grades function as effective demand-side signals, and when might their influence be weakened or undermined? 1.2. Reformulation pathways and a potential credibility problem In beverage markets, and especially in carbonated soft drinks (CSDs), reformulation to improve nutrient grades often relies on sugar reduction through sweetener substitution rather than through reductions in sweetness intensity or portion size [ 6 ] . Artificial sweeteners and sweetener blends are therefore a dominant pathway through which products can achieve more favorable algorithmic nutrient grades while maintaining taste profiles and competitive pricing. From a regulatory standpoint, such reformulation pathways are attractive because they enable relatively rapid improvements in nutrient profiles without major changes to production processes or consumer-facing characteristics. Yet this strategy may also introduce a potential credibility problem for nutrient grading systems [ 7 – 9 ] . While the grade communicates a simplified signal of "healthfulness" based on nutrient composition, ingredient lists—particularly those featuring artificial additives—can activate consumer beliefs related to processing, naturalness, and product authenticity. A growing body of literature suggests that many consumers interpret "naturalness" as an important heuristic for healthfulness, safety, and quality. Ingredient cues such as artificial sweeteners are frequently perceived as less natural and, in some cases, as undesirable, even when they contribute to reduced sugar content [ 5 , 8 , 10 ] . As a result, products that simultaneously display a favorable nutrient grade and contain additive-salient ingredients may present consumers with conflicting evaluative cues. This tension points to a potential limitation of nutrient grading policies: if the reformulation pathway used to achieve a high grade is perceived as illegitimate or inconsistent with consumer expectations, the grade itself may lose credibility [ 6 , 11 ] . In such cases, nutrient grades may not function as unconditional health signals, but rather as conditional cues whose influence depends on congruence with ingredient-level information. 1.3. The "clean-label paradox" and conditional effectiveness of nutrient grades We refer to this situation as a "clean-label paradox." Under this paradox, a product can achieve a high nutrient grade by reducing sugar through artificial sweeteners, yet the same reformulation strategy may undermine consumer acceptance by triggering skepticism toward additives and perceived artificiality. Rather than reinforcing the intended policy signal, the favorable grade may be discounted, ignored, or even contested by consumers. From a policy evaluation perspective, this paradox is consequential. Nutrient grading systems are typically assessed based on their average effects on purchasing behavior or nutrient intake [ 7 , 10 , 11 ] . However, if a substantial share of consumers systematically discounts high grades when they are paired with certain reformulation strategies, then the realized effectiveness of the policy may be lower than expected—even if compliance with the grading algorithm is high. The clean-label paradox also aligns with insights from cognitive dissonance theory. When consumers encounter information that implies inconsistent evaluations—such as a "healthy" grade alongside an ingredient profile perceived as artificial—they may experience psychological discomfort and resolve this inconsistency by reinterpreting or devaluing one of the signals [ 12 – 14 ] . In this context, downgrading the credibility or relevance of the nutrient grade represents a plausible resolution strategy. Despite its policy relevance, empirical evidence on how consumers integrate nutrient grades with ingredient cues remains limited. In particular, few studies have explicitly tested whether the positive valuation of favorable nutrient grades is contingent on the reformulation pathway used to achieve them, or whether such contingency translates into measurable economic outcomes such as willingness-to-pay. 1.4. Study objectives and contributions This study investigates whether and how the effectiveness of front-of-pack nutrient grades depends on ingredient-level reformulation strategies in the context of carbonated soft drinks [ 7 , 13 , 15 ] . Using a large-scale discrete choice experiment with Chinese consumers, we examine how consumers trade off algorithmic nutrient grades against sweetener strategies that differ in perceived naturalness, and whether favorable grades are discounted when paired with artificial sweeteners. Our analysis makes three contributions to the food policy literature. First, by modeling interactions between nutrient grades and sweetener strategies, we provide direct evidence on whether nutrient grades function as conditional policy signals rather than as unconditional health heuristics [ 16 ] . Second, by translating preference estimates into willingness-to-pay measures, we quantify the economic magnitude of any valuation erosion associated with grade–ingredient conflicts [ 17 – 18 ] . Third, through latent class analysis and post-choice psychological measures, we document heterogeneity in responses and identify consumer segments for whom the clean-label paradox is particularly salient [ 6 , 12 , 19 – 21 ] . Taken together, these contributions extend existing evaluations of front-of-pack labeling beyond average effects, highlighting the importance of reformulation pathways and consumer belief systems in shaping the realized impact of nutrient grading policies. 2. Materials and Methods 2.1. Study design and policy context This study tests whether front-of-pack (FOP) nutrient grades function as unconditional demand-side signals, or whether their influence is conditional on the reformulation pathway used to obtain favorable grades. The policy motivation is straightforward: in beverage markets, improved nutrient grades are often achieved through sugar reduction via sweetener substitution, including artificial sweeteners. If ingredient cues associated with such reformulation undermine the credibility of favorable grades, the realized effectiveness of nutrient grading policies may be attenuated. To quantify these trade-offs and test for cue conflict directly, we implemented a discrete choice experiment (DCE) grounded in random utility theory. The overall logic linking the policy mechanism, experimental identification, and econometric inference is summarized in Fig. 1 . 2.2. Sample, recruitment, and data quality control Participants were recruited via Wenjuanxing, a large mobile-based panel provider in China. To improve coverage relative to unconstrained opt-in sampling, quota-based recruitment was used to approximate census-informed distributions across gender, age, and city tier. Eligibility criteria targeted the active consumer base for carbonated soft drinks (CSDs): respondents were aged 18–45 and reported at least occasional CSD purchase or consumption within the past month. A total of 3,000 respondents completed the questionnaire. Data quality procedures were specified prior to analysis to ensure meaningful engagement with repeated choice tasks in a mobile environment. We excluded respondents with unusually rapid completion ("speeding"), those failing instructed-response attention checks, and those exhibiting patterned responding in non-choice batteries indicative of low effort. After applying these pre-specified screens, the final analytic sample consisted of 2,736 respondents. 2.3. Discrete choice experiment: attributes, levels, and choice task format The DCE was designed to approximate a stylized point-of-purchase decision for a single-serve CSD while maintaining strong internal validity. Each task presented two generic 330 mL can alternatives and a non-forced opt-out option ("I would buy neither"), reflecting real-world conditions in which consumers can reject all available options. Brand identifiers were intentionally omitted and the visual design of the can was held constant across tasks so that observed choice variation could be attributed to experimentally manipulated attributes rather than brand priors. Four attributes were included to map onto policy-relevant signals and industry reformulation strategies. First, the FOP nutrient grade served as the algorithmic health signal and varied across five levels (A, B, C, D, and no label). Second, sweetener strategy operationalized ingredient-based cues of naturalness versus artificiality via four formulations commonly used in the beverage sector: high fructose corn syrup (HFCS), cane sugar, an artificial sweetener blend (aspartame + acesulfame K), and a natural substitute blend (stevia + erythritol). Third, flavor (Classic, Citrus, Peach) was included to improve realism and reduce the likelihood that respondents based choices solely on evaluative cues (grade and sweetener). Fourth, price (3.0, 4.5, 6.0 RMB) was included to enable willingness-to-pay (WTP) estimation over a realistic single-serve price range. 2.4. Experimental design, blocking, and survey implementation A D-efficient fractional factorial design was generated in Ngene, producing 24 choice tasks. To limit respondent burden, tasks were blocked into two survey versions, and each participant completed 12 tasks. Block assignment was randomized at the respondent level, and the left–right placement of product profiles was randomized within tasks to mitigate position and ordering effects. Before the DCE, respondents read a brief neutral explanation stating that the nutrient grade is a front-of-pack summary derived from nutrient information. The wording avoided evaluative framing to reduce priming and to preserve the interpretation of grades as policy-relevant informational signals rather than normative endorsements. 2.5. Post-choice measures Immediately after completing the choice tasks, respondents answered a short post-choice module used to validate the intended cue structure and support interpretation of heterogeneity in choice behavior. Perceived naturalness ratings were collected for each sweetener strategy as a manipulation check. State cognitive dissonance was measured with items framed around respondents' immediate feelings about the trade-offs embedded in the purchase scenarios. Health literacy was assessed using a Chinese-adapted Newest Vital Sign instrument. These measures were used to contextualize segment differences and mechanism interpretation rather than to substitute for revealed preference evidence from the DCE. 2.6. Econometric Modeling and Statistical Analysis Choices were analyzed under RUT using a Random Parameters Logit (RPL) to accommodate preference heterogeneity and relax the IIA assumption. Utility for individual n , alternative j , and choice task t was specified as Eq. ( 1 ), $$\:{U}_{njt}\:=\:{V}_{njt}\:+\:{\epsilon\:}_{njt}$$ 1 with the systematic component modeled as a function of price, Nutri-Grade, sweetener strategy, flavor, and their interactions. Specifically, we estimated \(\:{V}_{njt}\) through Eq. ( 2 ), $$\:{V}_{njt}=\alpha\:{Price}_{njt}+{\beta\:}_{1}{Grade}_{njt}+{\beta\:}_{2}{Sweetener}_{njt}+{\beta\:}_{3}({Grade}_{njt}\times\:{Sweetener}_{njt})+\gamma\:{Flavor}_{njt}+{ASC}_{opt-out}$$ 2 where \(\:{ASC}_{opt-out}\) captures baseline propensity to reject both products. The key test of the clean-label paradox is the interaction between favorable grades—particularly Grade A—and the artificial sweetener blend (aspartame + acesulfame K). A significantly negative interaction term indicates that the positive utility of a high Nutri-Grade is systematically discounted when paired with additive-salient ingredients, consistent with the proposed dissonance-based mechanism [ 4 , 15 , 26 – 27 ] . Marginal willingness-to-pay (WTP) for attribute levels was calculated as Eq. ( 3 ), $$\:{WTP}_{k}\:=\:{-\beta\:}_{k}\:/\:\alpha\:$$ 3 with uncertainty quantified using the delta method or bootstrapping. To complement continuous heterogeneity in the RPL, we also estimated a Latent Class Model (LCM) to identify distinct preference segments, such as "Nutri-Grade followers" versus "clean-label purists," with the number of classes chosen based on information criteria and interpretability; segment profiles were then compared on consumption intensity and the post-choice psychological measures [ 17 ] . All analyzes were conducted in Python (PyBiogeme) and Stata 18.0, and robustness checks included alternative coding schemes (effects versus dummy coding), re-estimation excluding respondents flagged by quality screens, and comparisons across MNL, RPL, and LCM specifications. 2.7. Ethics Ethical review and approval were waived by the Institutional Review Board of Tianjin University of Traditional Chinese Medicine due to the anonymous and minimal-risk nature of the study. All participants provided electronic informed consent prior to participation. All methods were carried out in accordance with the relevant guidelines and regulations, and in accordance with the Declaration of Helsinki. 3. Results 3.1. Sample characteristics and data quality A total of 3,000 respondents completed the survey. After applying the pre-specified quality-control protocol (speeding, failed instructed-response attention checks, and patterned responding in psychometric batteries), 264 responses were excluded and the final analytic sample consisted of 2,736 respondents (retention rate 91.2%). The sample achieved a near-even gender split (49.1% male; 50.9% female) and broad geographic coverage across city tiers (Tier 1 to Tier 3 and below). As summarized in Table 1 , respondents were relatively engaged with the product category: 41.1% reported heavy CSD consumption (≥ 3 times/week), and 61.3% reported checking nutrition labels often or always. The age distribution aligned with quota targets (mean age 29.4 years, SD 6.8), and educational attainment skewed toward higher education, consistent with patterns commonly observed in opt-in digital panels (Table 1 ). Table 1 Sample characteristics and category engagement. Variable Category Frequency (n) Percentage (%) A. Socio-Demographics Gender Male 1,344 49.12% Female 1,392 50.88% Age Group 18–24 years 887 32.42% 25–34 years 1,113 40.68% 35–45 years 736 26.90% Marital Status Single / Unmarried 1,434 52.41% Married / Partnered 1,225 44.77% Divorced / Other 77 2.81% City Tier Tier 1 608 22.22% New Tier 1 724 26.46% Tier 2 775 28.33% Tier 3 and below 629 22.99% Education Level High school or below 312 11.40% Associate degree / College 588 21.49% Bachelor's degree 1,521 55.59% Master's degree or higher 315 11.51% Monthly Household Income < 10,000 RMB 843 30.81% 10,000–19,999 RMB 1,082 39.55% ≥ 20,000 RMB 811 29.64% B. Health Status & Lifestyle BMI Status * Underweight (18.5) 251 9.17% Normal weight (18.5–23.9) 1,678 61.33% Overweight/Obese (≥ 24.0) 807 29.50% Family History of Diabetes Yes 755 27.60% No / Unsure 1,981 72.40% Current Dieting Status On a diet 980 35.82% Not currently dieting 1,756 64.18% Self-Rated Health Excellent / Very Good 866 31.65% Good 1,352 49.42% Fair / Poor 518 18.93% C. CSD Category Engagement Consumption Frequency Light (1–3 times/month) 788 28.80% Moderate (1–2 times/week) 823 30.08% Heavy (≥ 3 times/week) 1,125 41.12% Primary Purchase Channel Convenience Store (C-Store) 1,768 64.62% Supermarket / Hypermarket 553 20.21% E-commerce / O2O Delivery 415 15.17% Label Reading Habit Never / Rarely 416 15.20% Sometimes 643 23.50% Often / Always 1,677 61.29% Note: Percentages may not sum to 100 due to rounding. City tier follows the Tier 1 / New Tier 1 / Tier 2 / Tier 3 and below classification used in quota sampling. "Heavy consumption" denotes carbonated soft drink (CSD) intake ≥ 3 times/week. BMI categories follow the Working Group on Obesity in China (WGOC) guideline cut-points. 3.2. Measurement properties and manipulation checks Before estimating choice models, we verified that post-choice constructs and experimental manipulations supported the intended interpretation of "ingredient naturalness cues" and mechanism-related measures. Confirmatory factor analysis indicated that the three-factor measurement model (State Cognitive Dissonance, Perceived Naturalness, and Food Tech Neophobia) fit the data well, with standard indices meeting conventional benchmarks; internal consistency and convergent validity were also strong, with Cronbach's α, composite reliability, and AVE values exceeding commonly used thresholds (see Table 2 ). Table 2 Psychometric properties of post-choice measures. Latent Construct and Items Std. Loading (λ) Cronbach's α CR AVE State Cognitive Dissonance (SCD) 0.893 0.893 0.732 SCD1: I felt conflicted choosing this drink because of its ingredients. 0.841 SCD2: I was uncomfortable with the trade-off between grade and additives. 0.882 SCD3: The choice left me feeling uneasy about the product's naturalness. 0.853 Perceived Naturalness (PN) 0.921 0.921 0.793 PN1: This product feels natural to me. 0.916 PN2: I perceive this product as free from artificial processing. 0.883 PN3: This product aligns with my idea of "clean label". 0.878 Food Tech Neophobia (FTN) 0.865 0.872 0.685 FTN1: Artificial sweeteners in food are unnecessary. 0.819 FTN2: Society is too dependent on food technology. 0.836 FTN3: New food additives are something I try to avoid. 0.843 Notes: Std. Loading = standardized factor loading from CFA; α = Cronbach's alpha; CR = composite reliability; AVE = average variance extracted. All loadings are statistically significant (p < 0.001). Crucially, the sweetener attribute generated a pronounced and theoretically consistent naturalness gradient. As shown in Fig. 2 , cane sugar was perceived as the most natural formulation (M = 5.82), followed by the stevia/erythritol blend (M = 5.43), with HFCS rated intermediate (M = 3.64). The artificial sweetener blend (aspartame + acesulfame K) was rated substantially less natural than all other conditions (M = 2.15), and Bonferroni-adjusted comparisons indicated that it differed significantly from every other sweetener strategy (all p < 0.001). This manipulation check confirms that the DCE operationalized ingredient-based "naturalness versus artificiality" cues in a way that is perceptually salient to respondents, establishing the necessary condition for testing whether nutrient-grade effects are contingent on reformulation strategy. 3.3. Baseline preference structure: main-effects random parameters logit We first estimated a random parameters logit (RPL) model with main effects to characterize baseline preference patterns. Full estimates are reported in Table 3 . The price coefficient was negative and statistically significant (β = −0.184, p < 0.001), indicating economically coherent choice behavior within the observed price range. The opt-out constant was large and negative (ASC = − 2.56, p < 0.001), consistent with a realistic tendency to reject both offered products in some tasks rather than being forced into a purchase. Table 3 Baseline preferences for nutrient grades and sweetener strategies: random parameters logit. Attribute Level Coef. (β) S.E. z-value p-value 95% Conf. Interval Price (RMB) — -0.184 0.042 -4.38 < 0.001 [-0.266, -0.102] Nutri-Grade Grade A 0.315 0.058 5.43 < 0.001 [ 0.201, 0.429] (Ref: Grade C) Grade B 0.072 0.05 1.45 0.145 [-0.025, 0.169] Grade D -0.248 0.056 -4.43 < 0.001 [-0.358, -0.138] None (No Label) -0.024 0.062 -0.39 0.698 [-0.146, 0.098] Sweetener Cane Sugar 0.295 0.052 5.67 < 0.001 [ 0.193, 0.397] (Ref: HFCS) Natural Blend (Stevia) 0.045 0.049 0.91 0.362 [-0.051, 0.141] Artificial Blend (Aspartame) -0.308 0.065 -4.74 < 0.001 [-0.435, -0.181] Flavor Citrus 0.015 0.041 0.37 0.714 [-0.065, 0.095] (Ref: Classic) Peach 0.068 0.04 1.7 0.089 [-0.010, 0.146] Constant Opt-out (ASC) -2.56 0.18 -14.22 < 0.001 [-2.913, -2.207] Random Parameters (SD) SD_Price 0.025 0.03 0.83 0.405 SD_Grade A 0.485 0.092 5.27 < 0.001 SD_Sweetener (Artificial) 1.34 0.115 11.65 < 0.001 Model Fit Statistics Observations 32,832 Log-Likelihood -26,104.50 McFadden Pseudo R 2 0.168 AIC 52,245.00 BIC 52,396.70 Notes: The dependent variable is the choice among two product alternatives plus an opt-out option ("buy neither"). Reference levels: Nutri-Grade = Grade C; Sweetener strategy = HFCS; Flavor = Classic. The price is in RMB. The opt-out term is an alternative-specific constant (ASC). Random-parameter standard deviations (SD) are reported for parameters specified as random. Bold indicates p < 0.05. Nutri-Grade effects exhibited a non-linear response pattern rather than a smooth monotonic gradient. Relative to Grade C (reference), Grade A generated a robust positive utility premium (β = 0.315, p < 0.001), while Grade B was small and not statistically distinguishable from zero (β = 0.072, p = 0.145). In contrast, Grade D was significantly penalized (β = −0.248, p < 0.001), and the "no label" condition was not significantly different from Grade C. This "threshold-like" pattern suggests that, in this category, consumers place disproportionate value on the top grade while treating intermediate grades as less salient. Sweetener strategy contributed independently to utility. Cane sugar was preferred over HFCS (β = 0.295, p < 0.001), whereas the artificial sweetener blend carried a significant disutility (β = −0.308, p < 0.001). The stevia/erythritol blend did not yield a statistically significant improvement relative to HFCS in the main-effects model (β = 0.045, p = 0.362). Importantly, the random-parameter estimates indicated substantial unobserved heterogeneity for Grade A and for the artificial sweetener blend, consistent with polarized responses toward artificial sweeteners. Flavor coefficients were comparatively small, implying that choices were driven primarily by the health–ingredient–price bundle rather than flavor variation. 3.4. Conditional effectiveness: grade × sweetener interactions ("clean-label paradox") To test whether the effect of nutrient grades depends on reformulation cues, we extended the RPL specification by including interactions between nutrient grade levels and sweetener strategies. Adding these interaction terms substantially improved model fit relative to the main-effects specification (likelihood-ratio test χ²(6) = 142.5, p < 0.001), indicating that consumers do not evaluate nutrient grades independently of ingredient composition. Key interaction estimates are reported in Table 4 . Table 4 Conditional effectiveness of nutrient grades: key grade × sweetener interaction effects in the RPL model. Parameter Coef. (β) S.E. z-value p-value Main Effects (Baseline) Nutri-Grade A 0.425 0.062 6.85 < 0.001 Nutri-Grade B 0.084 0.055 1.53 0.126 Artificial Blend (Main) -0.215 0.07 -3.07 0.002 Interaction Effects (The Paradox) Grade A × Artificial Blend -0.485 0.088 -5.51 < 0.001 Grade B × Artificial Blend -0.185 0.082 -2.26 0.024 Grade A × Natural Blend (Stevia) -0.052 0.114 -0.46 0.648 Grade B × Natural Blend (Stevia) -0.015 0.108 -0.14 0.889 Model Fit Log-Likelihood -25,985.20 Pseudo R 2 0.172 Notes: Estimates are from the interaction specification that adds cross-attribute terms between Nutri-Grade levels and sweetener strategies. The focal "clean-label paradox" test is the interaction between favorable grades (especially Grade A) and the artificial sweetener blend (aspartame + acesulfame K); a negative coefficient indicates discounting of the grade premium under ingredient conflict. The model controls for price, flavor, and main effects; only theoretically central interactions are displayed. Bold indicates p < 0.05. The central result is a strong negative interaction between Grade A and the artificial sweetener blend (β = −0.485, p < 0.001). This coefficient implies that the utility premium attached to the top nutrient grade is materially discounted when the product contains additive-salient artificial sweeteners. In practical terms, the Grade A signal does not "stack" additively on top of an artificial sweetener formulation: the interaction penalty offsets much of the Grade A main effect, producing only a minimal net gain from upgrading the grade under the artificial blend. By contrast, interactions between favorable grades and the stevia/erythritol blend were small and not statistically significant, suggesting that the discounting is not an inherent response to sugar reduction per se but is concentrated in the presence of artificial additives. This conditional pattern is also reflected in predicted choice probabilities. As illustrated in Fig. 3 , improvements in Nutri-Grade from C to A are associated with a clear rise in predicted choice probability for ingredient-congruent formulations (e.g., cane sugar), whereas the corresponding trajectory for the artificial sweetener blend is comparatively flat. This visualization reinforces the interpretation of nutrient grades as conditional signals: the demand response to a favorable grade is substantially weaker when ingredient cues conflict with the algorithmic health rating. 3.5. Willingness-to-pay: quantifying the valuation erosion To translate preference parameters into economically interpretable magnitudes, we computed marginal willingness-to-pay (WTP) for key contrasts using the ratio of attribute coefficients to the (negative) price coefficient. Results are summarized in Table 5 and visualized in Fig. 4 . Table 5 Willingness-to-pay contrasts quantify valuation erosion when Grade A is achieved via artificial sweeteners. Scenarios & Contrasts Estimated WTP (RMB) 95% Conf. Interval 1. The "Ideal" Health Premium Grade A vs. Grade C (with Cane Sugar) + 2.31 [1.85,2.77] 2. The "Paradox" Scenario Grade A vs. Grade C (with Artificial Blend) + 0.27 ns [− 0.15,0.69] 3. The "Paradox Penalty" (Diff) (Scenario 1) minus (Scenario 2) -2.04 [− 2.55,−1.53] 4. Reformulation Value Natural Blend (Stevia) vs. Artificial Blend + 1.68 [1.22,2.14] Note: Estimates based on the ratio of coefficients from the RPL interaction model. ns indicates the WTP estimate is not significantly different from zero at p < 0.05. In an ingredient-congruent context (Grade A relative to Grade C with cane sugar), respondents exhibited a sizable premium of + 2.31 RMB (95% CI [1.85, 2.77]), indicating that the top grade carries meaningful economic value when it aligns with expected ingredient cues. However, when the same Grade A upgrade occurs under the artificial sweetener blend, the implied premium collapses to + 0.27 RMB and is not statistically distinguishable from zero (95% CI [− 0.15, 0.69]). The difference between these scenarios implies a valuation erosion of approximately − 2.04 RMB (95% CI [− 2.55, − 1.53]), which quantifies the economic magnitude of the grade–ingredient conflict. Figure 4 presents these WTP contrasts directly, highlighting the sharp divergence in the monetary value of Grade A across reformulation contexts. WTP estimates also inform the valuation of alternative reformulation pathways. Replacing the artificial sweetener blend with the stevia/erythritol blend yields a positive WTP uplift of + 1.68 RMB (95% CI [1.22, 2.14]), consistent with a meaningful consumer premium for sugar-reduction strategies that are more compatible with clean-label interpretations. 3.6. Heterogeneity: latent class segmentation and interpretation Given the strong heterogeneity suggested by the random-parameters estimates, we estimated latent class models to identify discrete segments with qualitatively different evaluation rules. Model comparison favored a three-class solution balancing fit and interpretability. Segment sizes and defining preference parameters are reported in Table 6 , with psychometric profiling summarized in Fig. 5 . Table 6 Latent class segmentation reveals heterogeneous decision rules and identifies the segment driving grade discounting. Attributes Class 1: Clean-Label Purists Class 2: Algorithmic Health Seekers Class 3: Price-Sensitive Traditionalists Segment Size (Share) 44.20% 32.50% 23.30% Key Preferences (β) Price -0.12 -0.15 -0.65 *** Nutri-Grade A (Main) + 0.45 *** + 0.75 *** + 0.18 * Cane Sugar + 0.82 *** + 0.15 + 0.55 *** Artificial Blend (Main) -1.85 *** -0.15 ns -0.25 * The Paradox Interaction Grade A × Artificial -0.92 *** -0.08 ns -0.12 ns Interpretation "A-Grade is fake if artificial" "I trust the A-Grade" "Too expensive / I want real sugar" *** p < 0.001, * p < 0.05, ns not significant. Bold values indicate the defining characteristic of each class. The largest segment ("Clean-Label Purists", 44.2%) exhibited a pronounced aversion to the artificial sweetener blend (β = −1.85, p < 0.001) and the strongest discounting of Grade A when paired with artificial sweeteners (Grade A × artificial β = −0.92, p < 0.001). Consistent with the proposed cue-conflict mechanism, this segment also displayed higher post-choice state cognitive dissonance. A second segment ("Algorithmic Health Seekers", 32.5%) placed strong positive value on Grade A (β = 0.75, p < 0.001) while showing little evidence of interaction-based discounting and relatively weak sensitivity to artificial sweeteners, consistent with reliance on the front-of-pack grade as a primary heuristic. A third segment ("Price-Sensitive Traditionalists", 23.3%) was characterized by the strongest price sensitivity and a preference for cane sugar formulations, with comparatively muted responsiveness to nutrient grades. Together, these segments indicate that the clean-label paradox is not uniform across the population: it is driven primarily by a large segment that treats artificial sweeteners as a salient negative ingredient cue and correspondingly discounts favorable algorithmic grades. 3.7. Robustness checks Robustness checks indicate that the central inference—the negative Grade A × artificial sweetener interaction—does not depend on a single modeling choice. As shown in Appendix Table A1, the key interaction remains negative and statistically significant across MNL, RPL, and latent class specifications. Sensitivity analyses using effects coding yield substantively similar conclusions (Appendix Table A2), and re-estimation on stricter quality-screened subsamples produces closely aligned effect sizes and WTP penalties (Appendix Table A3). 4. Discussion 4.1. Main findings and why they matter for front-of-pack policy evaluation This study provides evidence that front-of-pack (FOP) nutrient grades in beverages operate as conditional signals rather than unconditional demand-side cues. In a discrete choice setting, respondents valued a favorable nutrient grade (especially Grade A), but the magnitude of that valuation depended critically on how the grade was achieved. When the grade improvement was paired with an artificial sweetener blend, the Grade A premium was sharply attenuated (Grade A × artificial blend interaction β = −0.485, p < 0.001), and the associated willingness-to-pay premium for upgrading from Grade C to Grade A effectively collapsed—from + 2.31 RMB under cane sugar to + 0.27 RMB (not statistically distinguishable from zero) under artificial sweeteners, implying a valuation erosion of approximately 2.04 RMB. From a Food Policy perspective, the key implication is not simply that "consumers dislike artificial sweeteners." Rather, the results speak to an instrument-design issue: the effectiveness of nutrient profiling policies depends on the coherence between the algorithmic signal that policy puts on the front of pack and the ingredient cues that consumers treat as relevant to healthfulness, legitimacy, or trust [ 15 , 21 , 22 ] . If a scoring system is implemented in a policy environment where manufacturers can most readily secure favorable grades via additive-salient reformulation pathways, then demand-side responses to the policy may be weaker than expected even when the policy is implemented correctly and information is technically accurate [ 23 – 25 ] . 4.2. Mechanism interpretation: credibility conflict as a behavioral constraint on policy impact The pattern observed here is consistent with a credibility-conflict mechanism: consumers appear to discount the informational value of a favorable grade when it is paired with ingredient cues interpreted as "artificial." The manipulation check confirms that the artificial blend is perceived as substantially less natural than alternative sweetening strategies, providing a credible basis for the cue-conflict interpretation [ 21 , 24 , 25 ] . This matters because most policy evaluations of nutrient profiling labels implicitly assume separability: a better grade should increase perceived healthfulness and choice probability regardless of how the product was reformulated. Our results challenge that separability assumption in a concrete way. The negative interaction indicates that the grade premium does not transfer across reformulation contexts; instead, the grade's marginal value is contingent on ingredient congruence [ 17 , 26 ] . In practice, this implies that nutrient grades can lose part of their "behavioral purchasing power" precisely in the situations where policy expects them to matter—when products are reformulated to obtain better scores. Notably, the attenuation is not a generic response to sugar reduction. Interactions involving the stevia/erythritol blend were not significant, suggesting that consumers do not automatically penalize "lower sugar" reformulation [ 11 , 14 , 27 – 28 ] . The discounting appears concentrated in reformulation pathways that activate salient beliefs about additives and artificial processing. 4.3. Reformulation pathways as a missing link in label policy effectiveness A central contribution of this study is to treat reformulation pathway choice as part of the policy mechanism rather than as an external industry detail. Nutrient-grade policies do not only inform consumers; they also shape manufacturer incentives by rewarding particular compositional changes. In beverages, the least costly and most technically straightforward pathway to improved nutrient scores is often substitution with non-nutritive sweeteners [ 13 , 16 , 24 , 29 ] . Our findings indicate that this pathway may generate a demand-side penalty that erodes the benefit of achieving a favorable grade. This suggests an important refinement to how nutrient profiling policies are conceptualized and evaluated: implementation success should not be assessed solely by uptake of labels or by the proportion of products achieving better grades. Instead, evaluation should also track the distribution of reformulation pathways used to obtain those grades and whether the demand response differs across those pathways [ 30 ] . If a policy disproportionately induces "score-improving" reformulation that is viewed skeptically by a large consumer segment, aggregate improvements in on-pack grades may overstate real-world shifts in purchasing and welfare. 4.4. Heterogeneity and distributional consequences: who follows the grade, and who contests it Average effects can obscure meaningful distributional patterns. The latent class results highlight three segments with different decision rules, and these segments are policy-relevant. A large segment of "clean-label purists" (44.2%) drives the paradox: they strongly dislike the artificial sweetener blend and exhibit the strongest within-segment discounting of Grade A when paired with artificial sweeteners. In contrast, "algorithmic health seekers" (32.5%) respond strongly to Grade A and do not show meaningful discounting, while "price-sensitive traditionalists" (23.3%) are driven primarily by price and prefer cane sugar with comparatively muted responsiveness to nutrient grades. This heterogeneity suggests that nutrient grading policies may produce uneven outcomes across groups even when the label is equally visible. If a sizable share of consumers systematically contests or discounts grades for certain reformulation strategies, then policy effects may be concentrated among those who treat the grade as a trusted heuristic, while other consumers remain anchored to ingredient-based evaluation [ 14 , 18 , 31 ] . This has practical implications for both equity and effectiveness: a "one-size-fits-all" informational tool may not generate uniform behavioral responses, and market adjustments (e.g., product offerings and pricing) may interact with these segment differences. 4.5. Policy implications: designing and evaluating nutrient grades under reformulation and contestation The findings point to a broader lesson for food labeling policy: front-of-pack nutrient grades are not merely informational devices; they are policy instruments that jointly shape consumer inference and firm behavior. In that sense, the key design problem is not "whether a grade is accurate," but whether the grade remains credible and behaviorally operative when firms pursue the most cost-effective pathways to improve it [ 13 , 17 , 32 – 33 ] . Our evidence indicates that the demand-side value of a top grade can be substantially discounted when the grade is produced through additive-salient reformulation (artificial sweetener substitution), with the economic signature of this discounting visible in the collapse of the Grade A premium in WTP terms. A useful way to translate this into policy design is to view nutrient grades as signals with endogenous production. The "signal" (Grade A) is generated by an algorithm, but the conditions under which products achieve the signal are shaped by industry incentives and constraints (cost, formulation feasibility, taste preservation). When the cheapest route to a stronger signal produces ingredient cues that many consumers interpret as inconsistent with "healthfulness," the policy can encounter a credibility constraint: the signal is present, yet its marginal persuasive power is reduced. This implies that the real-world effect of a grading system depends on (i) the scoring rule and threshold structure, (ii) the distribution of feasible reformulation options, and (iii) prevailing consumer beliefs about the legitimacy of those options. Three implications follow for policy design and implementation. First, reformulation pathways should be treated as a core outcome in policy evaluation, not a background condition. Many evaluations implicitly equate improved grades on shelves with improved public health impact. Our results caution against that equivalence: if grade improvements are disproportionately achieved through reformulation pathways that trigger skepticism, the policy may produce "paper gains" in grades with limited shifts in demand. A practical evaluation framework should therefore report, alongside grade distributions, the composition of reformulation strategies used to reach those grades (e.g., sugar reduction via sweeteners versus other adjustments) and test whether demand responses differ across those pathways. Second, the policy architecture may need complementary interpretive scaffolding to preserve the grade's meaning under cue conflict. The underlying issue is not that consumers are "irrational," but that they integrate multiple cues. If the grade is interpreted as a holistic endorsement while ingredient cues trigger "artificiality" concerns, conflict is predictable. Policy can respond by improving interpretability of what the grade represents (a nutrient-profile summary) and, importantly, what it does not represent. In practice, this does not require turning the front-of-pack into a complex label. It can involve standardized explanatory language in policy rollouts, consistent public communication, and—where appropriate—harmonization with other informational schemes that address processing-related concerns (so consumers are not forced to infer "processing" from ingredients while inferring "health" from grades). Third, distributional heterogeneity implies that a single-grade instrument can yield uneven behavioral gains and potentially uneven welfare effects. The latent classes indicate that a large segment ("clean-label purists") drives the discounting of favorable grades when paired with artificial sweeteners, whereas another segment appears to follow the grade as a primary heuristic. Policy evaluation that reports only average effects may therefore mischaracterize the mechanism: the policy works strongly for some consumers but is contested by others, particularly where reformulation relies on additives. This has two practical implications. On the demand side, it suggests that complementary communication may need to be targeted or tailored. On the supply side, it suggests that manufacturers' optimal responses to a grading policy may involve product-line differentiation (e.g., parallel offerings that appeal to grade-followers versus ingredient-first consumers), which can reshape market structure and pricing. Policymakers should anticipate such endogenous adjustment rather than assuming a uniform shift. Taken together, these implications emphasize that the effectiveness of nutrient grading policies depends on the interaction between algorithm design and industry response. The central policy takeaway is not that grades "fail," but that they can become conditional and contested when score-improving reformulation is achieved via ingredients that consumers treat as qualitatively meaningful. 4.6. Limitations and scope conditions Several limitations delineate the scope conditions of our inferences and point to where additional evidence is needed. First, discrete choice experiments approximate market decision-making but cannot fully capture dynamic purchasing and learning. Real consumers may update beliefs over repeated exposure (e.g., learning that a "Grade A + artificial sweeteners" product tastes good, or revising beliefs about sweeteners after information shocks). Our design captures preference trade-offs at the point of choice under controlled information, which is appropriate for isolating conditional signaling, but it does not model dynamic belief formation. Second, we intentionally used generic products to isolate policy-relevant cues; this strengthens internal validity but limits the ability to capture brand trust and reputation as moderators. In real markets, brand equity can either buffer skepticism (trusted brands may "carry" a grade further) or intensify it (brands associated with "naturalness" may face stronger backlash when using artificial additives). Whether brand moderates the grade–ingredient interaction is therefore an open empirical question. Third, external validity depends on the information environment. Perceptions of artificial sweeteners and "clean label" norms are shaped by media discourse, regulatory debates, and health communication. The magnitude of discounting is likely to be state-dependent: periods of heightened sweetener controversy or misinformation may amplify skepticism, whereas credible communication or shifting norms could attenuate it. Our results identify a robust behavioral pattern under the measured perceptions in our sample; translating effect sizes across contexts requires attention to those perception baselines. Fourth, the policy mechanism is inherently general-equilibrium: grading policies shape firm incentives, which can alter product portfolios, ingredient choices, and pricing. Our DCE holds many market features fixed to identify preference parameters, but in reality firms may respond strategically—e.g., by adjusting prices to offset demand penalties, changing package claims to reframe ingredient cues, or reallocating marketing budgets. These strategic responses could either mitigate or exacerbate the conditional effectiveness documented here. Recognizing this is important: the conditional signal is not just a consumer cognition issue; it is embedded in a market system where firms adapt. Finally, while the post-choice measures support a cue-conflict interpretation, they remain correlational with respect to mechanism. A more stringent mechanism test would experimentally manipulate interpretive frames (e.g., providing standardized explanations of what the grade reflects, or clarifying the safety/regulatory status of sweeteners) to test whether discounting is reduced without sacrificing label simplicity. 4.7. Future research The results motivate a research agenda that connects micro-level conditional signaling to macro-level policy performance. Field validation and repeated-choice behavior. A natural next step is to test whether the grade–ingredient interaction predicts realized purchasing in settings with real stakes—online grocery experiments, controlled shopping tasks with incentives, or retail scanner/panel data where available. Such studies could examine not only one-shot choice but also repeat purchase, substitution patterns across beverage categories, and persistence over time. Interpretability interventions. The credibility-conflict mechanism suggests that the effectiveness of nutrient grades may be improved by interventions that reduce misalignment between what the grade communicates and what consumers infer from ingredients. Future experiments could randomize brief, standardized interpretive statements (e.g., clarifying that the grade summarizes nutrients rather than processing), or compare alternative front-of-pack designs that better anticipate contested attributes. Importantly, the goal is to test whether interpretability can be increased without increasing cognitive load—an essential constraint for policy tools designed for rapid decision contexts. Endogenous reformulation and policy simulation. The most policy-relevant extension is to integrate demand estimates with a stylized supply response model. Because grading policies are likely to change the payoff to specific reformulation pathways, a complete evaluation should ideally combine (i) consumer demand response conditional on reformulation and (ii) firms' cost-minimizing reformulation choices. Even a simplified simulation—where firms choose between sugar reduction pathways with different cost and demand consequences—could illuminate when a grading policy is likely to induce "behaviorally robust" reformulation versus "behaviorally fragile" score improvements. This would move the literature closer to actionable ex ante policy assessment. Cross-category generalization. Finally, the conditional signal framework can be tested beyond beverages in categories where reformulation involves contested additives or processing cues (e.g., fat replacers, emulsifiers, preservatives). If conditional effectiveness is common, it strengthens the case for policy evaluation frameworks that explicitly treat algorithmic labels as signals whose credibility depends on how products achieve them. 5. Conclusions This study shows that front-of-pack nutrient grades in carbonated soft drinks function as conditional signals: favorable grades are valued, but their marginal influence is substantially reduced when achieved through additive-salient artificial sweetener reformulation. The resulting collapse of the Grade A WTP premium under artificial sweeteners quantifies the economic significance of this credibility conflict, while latent class results reveal that discounting is driven by a large consumer segment. For policy, the central message is that nutrient grading systems cannot be evaluated solely on label uptake or grade distributions. Their realized impact depends on the interaction between scoring rules and the reformulation pathways they induce, as well as on how consumers integrate algorithmic ratings with ingredient cues. Designing and evaluating nutrient grading policies with these behavioral and market-contingent mechanisms in mind is likely to be necessary for translating improved algorithmic scores into sustained shifts in purchasing and meaningful public health gains. Abbreviations The following abbreviations are used in this manuscript: AceK Acesulfame K (acesulfame potassium) AIC Akaike Information Criterion ANOVA Analysis of Variance ASC Alternative-Specific Constant AVE Average Variance Extracted AvePP Average Posterior Probability BIC Bayesian Information Criterion BMI Body Mass Index C-Store Convenience Store CFA Confirmatory Factor Analysis CFI Comparative Fit Index CI Confidence Interval CR Composite Reliability CSD(s) Carbonated Soft Drink(s) DCE Discrete Choice Experiment FOPL Front-of-Pack Labeling FTN Food Technology Neophobia HFCS High Fructose Corn Syrup IIA Independence of Irrelevant Alternatives IRB Institutional Review Board LCM Latent Class Model LR test Likelihood Ratio test MNL Multinomial Logit model MRS Marginal Rate of Substitution NVS Newest Vital Sign O2O Online-to-Offline PN Perceived Naturalness RMSEA Root Mean Square Error of Approximation RPL Random Parameters Logit (Mixed Logit) RUT Random Utility Theory SCD State Cognitive Dissonance SD Standard Deviation SE Standard Error SRMR Standardized Root Mean Square Residual TLI Tucker–Lewis Index WGOC Working Group on Obesity in China WTP Willingness to Pay Declarations Institutional Review Board Statement Ethical review and approval were waived for this study by the Institutional Review Board of Tianjin University of Traditional Chinese Medicine due to the anonymous nature of the data collection and the minimal risk posed to participants. Informed Consent Statement : Informed consent was obtained from all subjects involved in the study. Conflicts of Interest: The authors declare no conflicts of interest. Funding: This research received no external funding. Author Contribution Conceptualization, Y.C. and M.W.; methodology, Y.C. and T.Z.; software, Y.C. and H.G.; validation, H.G., Z.L. and K.M.; formal analysis, Y.C.; investigation, Y.C., H.G. and Z.L.; resources, T.Z. and M.W.; data curation, H.G. and Z.L.; writing—original draft preparation, Y.C.; writing—review and editing, K.M., M.W. and T.Z.; visualization, Y.C. and Z.L.; supervision, T.Z. and M.W.; project administration, T.Z. and M.W.; funding acquisition, T.Z. and M.W. All authors have read and agreed to the published version of the manuscript. Acknowledgments: We would like to express our gratitude to the operational team at Wenjuanxing for their technical support in data collection and panel management. We also thank all the anonymous participants who took part in this survey. During the preparation of this manuscript, the authors used Gemini 3 for the purpose of English language editing and proofreading. The authors have reviewed and edited the output and take full responsibility for the content of this publication. Data Availability The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical considerations regarding the participants. Disclaimer/ Publisher's Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. References Villaverde, P. et al. Hypothetical impact of the Mexican front-of-pack labeling on intake of critical nutrients and energy. J. Health Popul. Nutr. 42 (1), 124. 10.1186/s41043-023-00462-7 (2023). PMID: 37941048; PMCID: PMC10631037. Bassetti, E., Khosravi, A. & Pries, A. M. Prevalence of Front-of-Pack Warning Signs among Commercial Complementary Foods in Seven High and Upper Middle-Income Countries. Nutrients 15 (7), 1629. 10.3390/nu15071629 (2023). PMID: 37049470; PMCID: PMC10096843. Roark, E. K., Rehm, C. D. & Sherry, C. L. Potential for Front of Pack Labeling Exposure to Impact US Dietary Choices: A Population-Based Cross-Sectional Study Using NHANES 2017–2018. Nutrients 14 (14), 2995. 10.3390/nu14142995 (2022). PMID: 35889951; PMCID: PMC9316916. Sarda, B. et al. Complementarity between the updated version of the front-of-pack nutrition label Nutri-Score and the food-processing NOVA classification. Public. Health Nutr. 27 (1), e63 (2024). PMID: 38297466; PMCID: PMC10897572. Mulligan, C., Lee, J. J., Vergeer, L., Ahmed, M. & L'Abbé, M. R. Evaluating the potential implications of canadian front-of-pack labelling regulations in generic and branded food composition databases. BMC Public. Health . 22 (1), 1866. 10.1186/s12889-022-14269-4 (2022). PMID: 36203128; PMCID: PMC9535871. Shin, S. et al. A Randomized Controlled Study to Test Front-of-Pack (FOP) Nutrition Labels in the Kingdom of Saudi Arabia. Nutrients 15 (13), 2904. 10.3390/nu15132904 (2023). PMID: 37447230; PMCID: PMC10343722. Lee, J. J. et al. Examining the diet quality of Canadian adults and the alignment of Canadian front-of-pack labelling regulations with other front-of-pack labelling systems and dietary guidelines. Front Public Health. ;11:1168745. doi: 10.3389/fpubh.2023.1168745. Erratum in: Front Public Health. 2024;12:1448853. (2023). 10.3389/fpubh.2024.1448853 . PMID: 37427256; PMCID: PMC10326271. Shin, S., Puri, J. & Finkelstein, E. A randomized trial to evaluate the impact of Singapore's forthcoming Nutri-grade front-of-pack beverage label on food and beverage purchases. Int. J. Behav. Nutr. Phys. Act. 20 (1), 18. 10.1186/s12966-023-01422-4 (2023). PMID: 36793058; PMCID: PMC9930047. Sarda, B. et al. Consistency of the Initial and Updated Version of the Nutri-Score with Food-Based Dietary Guidelines: A French Perspective. J. Nutr. 154 (3), 1027–1038 (2024). Epub 2024 Feb 2. PMID: 38311063; PMCID: PMC10942857. Miller, C., Kay, E., Dono, J. & Ettridge, K. Impacts of sugar and sweetener warning labels on substitution between sugar- and non-sugar-sweetened beverages in a non-hypothetical selection task. BMC Med. 22 (1), 541. 10.1186/s12916-024-03740-1 (2024). PMID: 39558314; PMCID: PMC11574984. Miller, C., Ettridge, K., Kay, E. & Dono, J. What about 100% juice and non-sugar sweeteners? A national study of support for taxes, labelling and marketing bans applied to sugary drinks, non-sugar sweetened beverages and 100% juice in Australia. Aust N Z. J. Public. Health . 49 (3), 100238 (2025). Epub 2025 Apr 29. PMID: 40307103. Miller, C. et al. Warning labels for sugar-sweetened beverages and fruit juice: evaluation of 27 different labels on health effects, sugar content, energy and exercise equivalency. Public. Health . 230 , 138–148 (2024). Epub 2024 Mar 27. PMID: 38547760. Bassetti, E. et al. Benchmarking the nutrient composition and labelling practices of commercially produced ready-to-eat purées and meals for older infants and young children across seven Southeast Asian countries. Matern Child. Nutr. 19 (Suppl 2(Suppl 2), e13585. 10.1111/mcn.13585 (2023). PMID: 38092377; PMCID: PMC10719058. Sagaceta-Mejia, J. et al. Purchase behavior reported by Mexican adults after implementing the front-of-pack warning labels. Salud Publica Mex . 67 (5 (sept-oct)), 447–454 (2025). Spanish. doi: 10.21149/16253. PMID: 41150992. Boukid, F. & Castellari, M. Food and Beverages Containing Algae and Derived Ingredients Launched in the Market from 2015 to 2019: A Front-of-Pack Labeling Perspective with a Special Focus on Spain. Foods 10 (1), 173. 10.3390/foods10010173 (2021). PMID: 33467009; PMCID: PMC7830845. Kroker-Lobos, M. F. et al. Efficacy of front-of-pack warning label system versus guideline for daily amount on healthfulness perception, purchase intention and objective understanding of nutrient content of food products in Guatemala: a cross-over cluster randomized controlled experiment. Arch. Public. Health . 81 (1), 108. 10.1186/s13690-023-01124-0 (2023). PMID: 37328782; PMCID: PMC10273755. Quevedo, K. L. et al. Sociodemographic differences in nutrition labels effect on Chilean and Mexican youth. Rev. Panam. Salud Publica . 48 , e33 (2024). PMID: 38576841; PMCID: PMC10993807. Hashem, K. Towards a more nuanced understanding of policies that lead to food reformulation for a food system change. Proc Nutr Soc. 2025 Dec 26:1–23. doi: 10.1017/S002966512510205X. Epub ahead of print. PMID: 41449698. Tomaz, L. A. et al. From the most to the least flexible nutritional profile: Classification of foods marketed in Brazil according to the Brazilian and Mexican models. Front. Nutr. 9 , 919582 (2022). PMID: 36204372; PMCID: PMC9531871. Miller, C. et al. Qualitative insights into Australian consumers' views for and against government action on sugary drinks. Public Health Res Pract. ;31(2):30122003. (2021). 10.17061/phrp30122003 . PMID: 34104932. Crosbie, E. et al. A policy study on front-of-pack nutrition labeling in the Americas: emerging developments and outcomes. Lancet Reg. Health Am. 18 , 100400 (2022). PMID: 36844016; PMCID: PMC9950544. Martini, D. et al. Can Front-of-Pack Labeling Encourage Food Reformulation? A Cross-Sectional Study on Packaged Bread. Foods 13 (22), 3535. 10.3390/foods13223535 (2024). PMID: 39593951; PMCID: PMC11593041. Mialon, M. Front-of-pack nutrition labeling: ultra-processed foods are the elephant in the room. Lancet Reg Health Am. ;18:100442. (2023). 10.1016/j.lana.2023.100442 . PMID: 36844010; PMCID: PMC9950663. Devaux, M., Aldea, A., Lerouge, A., Vuik, S. & Cecchini, M. Establishing an EU-wide front-of-pack nutrition label: Review of options and model-based evaluation. Obes. Rev. 25 (6), e13719. 10.1111/obr.13719 (2024). Epub 2024 Feb 7. PMID: 38326224. Julia, C., Gokani, N., Hercberg, S. & Garde, A. Ten years of Nutri-Score front-of-pack nutrition labelling in Europe. Nat. Food . 6 (3), 239–243. 10.1038/s43016-025-01141-y (2025). Epub 2025 Mar 13. PMID: 40082656. Merz, B. et al. Nutri-Score 2023 update. Nat. Food . 5 (2), 102–110. 10.1038/s43016-024-00920-3 (2024). Epub 2024 Feb 14. PMID: 38356074. Kelly, B., Ng, S. H., Carrad, A. & Pettigrew, S. The Potential Effectiveness of Front-of-Pack Nutrition Labeling for Improving Population Diets. Annu. Rev. Nutr. 44 (1), 405–440. 10.1146/annurev-nutr-011224-030917 (2024). Epub 2024 Aug 12. PMID: 38857543. Yin, X. et al. Stakeholder Network Analysis for Front-of-Pack Labeling in China. Front. Nutr. 9 , 871062 (2022). PMID: 35677544; PMCID: PMC9169156. Song, J. et al. Impact of color-coded and warning nutrition labelling schemes: A systematic review and network meta-analysis. PLoS Med. 18 (10), e1003765. 10.1371/journal.pmed.1003765 (2021). PMID: 34610024; PMCID: PMC8491916. Fialon, M., Nabec, L. & Julia, C. Legitimacy of Front-of-Pack Nutrition Labels: Controversy Over the Deployment of the Nutri-Score in Italy. Int. J. Health Policy Manag . 11 (11), 2574–2587. 10.34172/ijhpm.2022.6127 (2022). Epub 2022 Feb 20. PMID: 35219283; PMCID: PMC9818094. Wu, F. et al. Prepackaged foods healthiness ranking index: developing front-of-pack labeling to facilitate the assessment and validation of healthiness rankings of prepackaged foods in China. BMC Med. 22 (1), 402. 10.1186/s12916-024-03600-y (2024). PMID: 39300429; PMCID: PMC11414037. Pettigrew, S. et al. A review of front-of-pack nutrition labelling in Southeast Asia: Industry interference, lessons learned, and future directions. Lancet Reg. Health Southeast. Asia . 3 , 100017 (2022). PMID: 37384259; PMCID: PMC10305914. Chung, A., Hatzikiriakidis, K., Martino, F. & Skouteris, H. Characterising Parent-Appeal Marketing on Foods for Children: A Scoping Review. Curr. Nutr. Rep. 13 (3), 393–398. 10.1007/s13668-024-00559-3 (2024). Epub 2024 Jun 27. PMID: 38935250; PMCID: PMC11327212. Additional Declarations No competing interests reported. Supplementary Files AppendixA.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8573773","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":577446721,"identity":"3c337ba2-91cc-465c-854f-74fce10f3d50","order_by":0,"name":"Yirui Chen","email":"","orcid":"","institution":"School of Public Health and Health Sciences, Tianjin University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yirui","middleName":"","lastName":"Chen","suffix":""},{"id":577446722,"identity":"8a5ef391-1adb-4ea9-9071-5873b7653a36","order_by":1,"name":"Hongxin Gui","email":"","orcid":"","institution":"School of Public Health and Health Sciences, Tianjin University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Hongxin","middleName":"","lastName":"Gui","suffix":""},{"id":577446723,"identity":"1c83f3bf-caa2-46ec-9e82-d63cbc7b7692","order_by":2,"name":"Zhiyuan Liu","email":"","orcid":"","institution":"School of Public Health and Health Sciences, Tianjin University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Zhiyuan","middleName":"","lastName":"Liu","suffix":""},{"id":577446724,"identity":"08588f26-593b-4d65-bbfa-3474562a9007","order_by":3,"name":"Kai Ma","email":"","orcid":"","institution":"School of Culture and Communication, Tianjin University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Kai","middleName":"","lastName":"Ma","suffix":""},{"id":577446725,"identity":"204f094f-8885-413a-a4cd-421bcbe00828","order_by":4,"name":"Tieniu Zhao","email":"","orcid":"","institution":"School of Public Health and Health Sciences, Tianjin University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Tieniu","middleName":"","lastName":"Zhao","suffix":""},{"id":577446726,"identity":"1f94224b-5d55-4b8d-a57e-6070e19ae228","order_by":5,"name":"Mengyang Wang","email":"data:image/png;base64,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","orcid":"","institution":"School of Public Health and Health Sciences, Tianjin University of Traditional Chinese Medicine","correspondingAuthor":true,"prefix":"","firstName":"Mengyang","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2026-01-11 13:23:25","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8573773/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8573773/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100761524,"identity":"caaa35ac-de17-43c0-b7e0-403754edefc1","added_by":"auto","created_at":"2026-01-21 07:46:00","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2956091,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript20260113.docx","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/ddcec930383fd78afafd2f8f.docx"},{"id":100761540,"identity":"a90ab28d-d1ad-4a60-aa04-6770c824de18","added_by":"auto","created_at":"2026-01-21 07:46:44","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7971,"visible":true,"origin":"","legend":"","description":"","filename":"d18841a46d2949979e8e6ea5fb3ecd63.json","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/56561b398cf86dfb8e3c9480.json"},{"id":100761509,"identity":"dfed46a9-bf2a-4a14-87c4-f1d510968177","added_by":"auto","created_at":"2026-01-21 07:45:41","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":187796,"visible":true,"origin":"","legend":"","description":"","filename":"d18841a46d2949979e8e6ea5fb3ecd631enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/5227ecb2f8e7437780b8ff0f.xml"},{"id":100761544,"identity":"f59c8faa-4595-43b1-a740-aa8f88713bec","added_by":"auto","created_at":"2026-01-21 07:47:01","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":145352,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/fbb767b2cf314c8481e73bbe.png"},{"id":100761539,"identity":"83319315-0335-4265-aa12-474c6721b09e","added_by":"auto","created_at":"2026-01-21 07:46:44","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":25959,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/87af3d514ab3b7d2fb7ec506.png"},{"id":100761523,"identity":"64c865c8-6c02-4e3f-a836-3a794bced36f","added_by":"auto","created_at":"2026-01-21 07:46:00","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36969,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/4cf386f673444c260fe77ebb.png"},{"id":100761506,"identity":"0f8df5a7-2209-4650-901e-c8de857ea4d5","added_by":"auto","created_at":"2026-01-21 07:45:35","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":24435,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/df356502a61c80bba974cee4.png"},{"id":100761507,"identity":"facb9812-ed7c-41c4-8153-53653d72b839","added_by":"auto","created_at":"2026-01-21 07:45:39","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":269980,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/3214da30fe2641132fc54e36.png"},{"id":100761535,"identity":"93f5a288-173b-470b-9780-65e7625ac13e","added_by":"auto","created_at":"2026-01-21 07:46:31","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":185232,"visible":true,"origin":"","legend":"","description":"","filename":"d18841a46d2949979e8e6ea5fb3ecd631structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/e4759b5a0dae08265d9fb075.xml"},{"id":100761555,"identity":"b95cc3d2-10dc-4e4e-83b7-ae7a705dfcfd","added_by":"auto","created_at":"2026-01-21 07:47:23","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":201059,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/e010e085eaa5d1b66d36e3c8.html"},{"id":100761532,"identity":"d770c94f-33b5-424a-8002-a4282ee6db83","added_by":"auto","created_at":"2026-01-21 07:46:21","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1102160,"visible":true,"origin":"","legend":"\u003cp\u003ePolicy mechanism, experimental identification, and econometric strategy for testing conditional responses to front-of-pack nutrient grades. The figure summarizes the study logic linking front-of-pack nutrient grades (as an algorithmic health signal) and reformulation pathways (sweetener substitution) to consumer choice. The discrete choice experiment jointly varies nutrient grades, sweetener strategies, and prices in a controlled purchase setting with an opt-out option, enabling estimation of main effects and grade × sweetener interactions that capture signal discounting under ingredient conflict. Post-choice measures of perceived naturalness and state cognitive dissonance validate the cue structure and inform interpretation of preference heterogeneity. Choice data are analyzed under random utility theory using random parameters logit and latent class models, and key contrasts are translated into willingness-to-pay to support policy-relevant interpretation.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/996ce14c34ba541116e58f73.png"},{"id":100761503,"identity":"65dc723c-9263-48ea-9724-19ba7744ff95","added_by":"auto","created_at":"2026-01-21 07:45:19","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":70225,"visible":true,"origin":"","legend":"\u003cp\u003eManipulation check: perceived naturalness differs sharply across sweetener strategies. Distributions of perceived naturalness ratings for each sweetener strategy (1–7 scale) are shown; mean estimates with 95% confidence intervals are overlaid. Cane sugar and the stevia/erythritol blend are rated substantially more natural than the artificial sweetener blend (aspartame + acesulfame K). Pairwise differences are Bonferroni-adjusted (*** p \u0026lt; 0.001).\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/604a4e999fb7ebc0c24f4b4c.png"},{"id":100761533,"identity":"37a9d2d9-fc3e-4f63-81d1-2c5913ac73a4","added_by":"auto","created_at":"2026-01-21 07:46:27","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":102696,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted choice probabilities show that Grade improvements have weaker effects under artificial sweeteners (grade × sweetener interaction). Predicted probabilities are generated from the RPL interaction model. The x-axis shows Nutri-Grade levels (with Grade C as baseline); the y-axis shows the probability of choosing the product. Shaded bands denote 95% confidence intervals. The relatively flat trajectory for the artificial sweetener blend indicates discounting of favorable grades under ingredient conflict.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/8f99cb5e449d537b727c5e69.png"},{"id":100761554,"identity":"77c51a65-a385-4c01-a8c9-d6a18ea06a86","added_by":"auto","created_at":"2026-01-21 07:47:22","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":74647,"visible":true,"origin":"","legend":"\u003cp\u003eWillingness-to-pay for Grade A depends on sweetener strategy; the Grade A premium collapses under artificial sweeteners. Forest plot of WTP estimates (RMB) for Grade A relative to Grade C under different sweetener contexts, derived from the RPL interaction model. Points denote mean WTP; horizontal bars denote 95% confidence intervals. Estimates that cross zero are not statistically distinguishable from zero at p \u0026lt; 0.05.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/a538983066c61dba3fc5a58a.png"},{"id":100761508,"identity":"7162787a-5ab2-4b36-9f81-9907d15f784f","added_by":"auto","created_at":"2026-01-21 07:45:41","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1542841,"visible":true,"origin":"","legend":"\u003cp\u003eSegment profiling: \"clean-label purists\" exhibit higher dissonance and stronger aversion to artificial sweeteners. The figure reports standardized (or mean) post-choice construct scores by latent class, including state cognitive dissonance, food technology neophobia, and health literacy (specify scaling as displayed). Segment definitions correspond to the three-class LCM in Table 6; higher dissonance and stronger neophobia in the \"clean-label purist\" class are consistent with ingredient-driven discounting of favorable nutrient grades.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/3a4024ae5f79a16f9a0aad6f.png"},{"id":103480762,"identity":"7294a4e9-d772-4924-9628-1fc2908b7e62","added_by":"auto","created_at":"2026-02-26 07:57:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4555047,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/4e419b3a-b64a-4176-b520-8e54b27be881.pdf"},{"id":100761522,"identity":"33e296bd-6561-4c69-8be9-aacea5c8ca91","added_by":"auto","created_at":"2026-01-21 07:45:59","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":19611,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixA.docx","url":"https://assets-eu.researchsquare.com/files/rs-8573773/v1/205b6c54037171eead5d3240.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Reformulation pathways shape the impact of front-of-pack nutrient grades: A discrete choice experiment on sweetener substitution","fulltext":[{"header":"1. Introduction","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e \u003ch2\u003e1.1. Front-of-pack nutrient grading as a policy instrument\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eExcess intake of free sugars is widely recognized as a major modifiable risk factor for obesity, type 2 diabetes, and other non-communicable diseases. Sugar-sweetened beverages are among the most prominent contributors to added sugar intake and, as a result, have become a central target of nutrition-related public policy interventions. In recent years, front-of-pack labeling (FOPL) systems\u0026mdash;particularly nutrient grading or scoring schemes\u0026mdash;have been increasingly adopted or proposed as policy tools to improve dietary choices and to incentivize product reformulation.\u003c/p\u003e \u003cp\u003eNutrient grading systems aim to simplify complex nutritional information into salient, easy-to-interpret signals that can guide consumer choice at the point of purchase. By translating nutrient profiles into summary grades or scores, these systems are expected to operate through two complementary channels: first, by shifting consumer demand toward products with more favorable grades; and second, by encouraging manufacturers to reformulate products in order to achieve higher grades\u003csup\u003e[\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. From a policy perspective, the effectiveness of nutrient grading thus depends not only on consumers' responsiveness to the label itself, but also on how firms strategically adapt their product formulations in response to the scoring criteria.\u003c/p\u003e \u003cp\u003eEmpirical evidence suggests that front-of-pack nutrient labels can influence purchasing behavior and improve the nutritional quality of food environments. However, growing research also indicates that the impact of these labels is heterogeneous across products, consumer groups, and implementation contexts\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e. This raises a critical policy question: under what conditions do nutrient grades function as effective demand-side signals, and when might their influence be weakened or undermined?\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e1.2. Reformulation pathways and a potential credibility problem\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eIn beverage markets, and especially in carbonated soft drinks (CSDs), reformulation to improve nutrient grades often relies on sugar reduction through sweetener substitution rather than through reductions in sweetness intensity or portion size\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. Artificial sweeteners and sweetener blends are therefore a dominant pathway through which products can achieve more favorable algorithmic nutrient grades while maintaining taste profiles and competitive pricing.\u003c/p\u003e \u003cp\u003eFrom a regulatory standpoint, such reformulation pathways are attractive because they enable relatively rapid improvements in nutrient profiles without major changes to production processes or consumer-facing characteristics. Yet this strategy may also introduce a potential credibility problem for nutrient grading systems\u003csup\u003e[\u003cspan additionalcitationids=\"CR8\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e. While the grade communicates a simplified signal of \"healthfulness\" based on nutrient composition, ingredient lists\u0026mdash;particularly those featuring artificial additives\u0026mdash;can activate consumer beliefs related to processing, naturalness, and product authenticity.\u003c/p\u003e \u003cp\u003eA growing body of literature suggests that many consumers interpret \"naturalness\" as an important heuristic for healthfulness, safety, and quality. Ingredient cues such as artificial sweeteners are frequently perceived as less natural and, in some cases, as undesirable, even when they contribute to reduced sugar content\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e. As a result, products that simultaneously display a favorable nutrient grade and contain additive-salient ingredients may present consumers with conflicting evaluative cues.\u003c/p\u003e \u003cp\u003eThis tension points to a potential limitation of nutrient grading policies: if the reformulation pathway used to achieve a high grade is perceived as illegitimate or inconsistent with consumer expectations, the grade itself may lose credibility\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. In such cases, nutrient grades may not function as unconditional health signals, but rather as conditional cues whose influence depends on congruence with ingredient-level information.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e1.3. The \"clean-label paradox\" and conditional effectiveness of nutrient grades\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eWe refer to this situation as a \"clean-label paradox.\" Under this paradox, a product can achieve a high nutrient grade by reducing sugar through artificial sweeteners, yet the same reformulation strategy may undermine consumer acceptance by triggering skepticism toward additives and perceived artificiality. Rather than reinforcing the intended policy signal, the favorable grade may be discounted, ignored, or even contested by consumers.\u003c/p\u003e \u003cp\u003eFrom a policy evaluation perspective, this paradox is consequential. Nutrient grading systems are typically assessed based on their average effects on purchasing behavior or nutrient intake\u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. However, if a substantial share of consumers systematically discounts high grades when they are paired with certain reformulation strategies, then the realized effectiveness of the policy may be lower than expected\u0026mdash;even if compliance with the grading algorithm is high.\u003c/p\u003e \u003cp\u003eThe clean-label paradox also aligns with insights from cognitive dissonance theory. When consumers encounter information that implies inconsistent evaluations\u0026mdash;such as a \"healthy\" grade alongside an ingredient profile perceived as artificial\u0026mdash;they may experience psychological discomfort and resolve this inconsistency by reinterpreting or devaluing one of the signals\u003csup\u003e[\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e. In this context, downgrading the credibility or relevance of the nutrient grade represents a plausible resolution strategy.\u003c/p\u003e \u003cp\u003eDespite its policy relevance, empirical evidence on how consumers integrate nutrient grades with ingredient cues remains limited. In particular, few studies have explicitly tested whether the positive valuation of favorable nutrient grades is contingent on the reformulation pathway used to achieve them, or whether such contingency translates into measurable economic outcomes such as willingness-to-pay.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e1.4. Study objectives and contributions\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThis study investigates whether and how the effectiveness of front-of-pack nutrient grades depends on ingredient-level reformulation strategies in the context of carbonated soft drinks\u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. Using a large-scale discrete choice experiment with Chinese consumers, we examine how consumers trade off algorithmic nutrient grades against sweetener strategies that differ in perceived naturalness, and whether favorable grades are discounted when paired with artificial sweeteners.\u003c/p\u003e \u003cp\u003eOur analysis makes three contributions to the food policy literature. First, by modeling interactions between nutrient grades and sweetener strategies, we provide direct evidence on whether nutrient grades function as conditional policy signals rather than as unconditional health heuristics\u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e. Second, by translating preference estimates into willingness-to-pay measures, we quantify the economic magnitude of any valuation erosion associated with grade\u0026ndash;ingredient conflicts\u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e. Third, through latent class analysis and post-choice psychological measures, we document heterogeneity in responses and identify consumer segments for whom the clean-label paradox is particularly salient\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTaken together, these contributions extend existing evaluations of front-of-pack labeling beyond average effects, highlighting the importance of reformulation pathways and consumer belief systems in shaping the realized impact of nutrient grading policies.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Study design and policy context\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThis study tests whether front-of-pack (FOP) nutrient grades function as unconditional demand-side signals, or whether their influence is conditional on the reformulation pathway used to obtain favorable grades. The policy motivation is straightforward: in beverage markets, improved nutrient grades are often achieved through sugar reduction via sweetener substitution, including artificial sweeteners. If ingredient cues associated with such reformulation undermine the credibility of favorable grades, the realized effectiveness of nutrient grading policies may be attenuated. To quantify these trade-offs and test for cue conflict directly, we implemented a discrete choice experiment (DCE) grounded in random utility theory. The overall logic linking the policy mechanism, experimental identification, and econometric inference is summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Sample, recruitment, and data quality control\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eParticipants were recruited via Wenjuanxing, a large mobile-based panel provider in China. To improve coverage relative to unconstrained opt-in sampling, quota-based recruitment was used to approximate census-informed distributions across gender, age, and city tier. Eligibility criteria targeted the active consumer base for carbonated soft drinks (CSDs): respondents were aged 18\u0026ndash;45 and reported at least occasional CSD purchase or consumption within the past month. A total of 3,000 respondents completed the questionnaire.\u003c/p\u003e \u003cp\u003eData quality procedures were specified prior to analysis to ensure meaningful engagement with repeated choice tasks in a mobile environment. We excluded respondents with unusually rapid completion (\"speeding\"), those failing instructed-response attention checks, and those exhibiting patterned responding in non-choice batteries indicative of low effort. After applying these pre-specified screens, the final analytic sample consisted of 2,736 respondents.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Discrete choice experiment: attributes, levels, and choice task format\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe DCE was designed to approximate a stylized point-of-purchase decision for a single-serve CSD while maintaining strong internal validity. Each task presented two generic 330 mL can alternatives and a non-forced opt-out option (\"I would buy neither\"), reflecting real-world conditions in which consumers can reject all available options. Brand identifiers were intentionally omitted and the visual design of the can was held constant across tasks so that observed choice variation could be attributed to experimentally manipulated attributes rather than brand priors.\u003c/p\u003e \u003cp\u003eFour attributes were included to map onto policy-relevant signals and industry reformulation strategies. First, the FOP nutrient grade served as the algorithmic health signal and varied across five levels (A, B, C, D, and no label). Second, sweetener strategy operationalized ingredient-based cues of naturalness versus artificiality via four formulations commonly used in the beverage sector: high fructose corn syrup (HFCS), cane sugar, an artificial sweetener blend (aspartame\u0026thinsp;+\u0026thinsp;acesulfame K), and a natural substitute blend (stevia\u0026thinsp;+\u0026thinsp;erythritol). Third, flavor (Classic, Citrus, Peach) was included to improve realism and reduce the likelihood that respondents based choices solely on evaluative cues (grade and sweetener). Fourth, price (3.0, 4.5, 6.0 RMB) was included to enable willingness-to-pay (WTP) estimation over a realistic single-serve price range.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Experimental design, blocking, and survey implementation\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eA D-efficient fractional factorial design was generated in Ngene, producing 24 choice tasks. To limit respondent burden, tasks were blocked into two survey versions, and each participant completed 12 tasks. Block assignment was randomized at the respondent level, and the left\u0026ndash;right placement of product profiles was randomized within tasks to mitigate position and ordering effects.\u003c/p\u003e \u003cp\u003eBefore the DCE, respondents read a brief neutral explanation stating that the nutrient grade is a front-of-pack summary derived from nutrient information. The wording avoided evaluative framing to reduce priming and to preserve the interpretation of grades as policy-relevant informational signals rather than normative endorsements.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Post-choice measures\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eImmediately after completing the choice tasks, respondents answered a short post-choice module used to validate the intended cue structure and support interpretation of heterogeneity in choice behavior. Perceived naturalness ratings were collected for each sweetener strategy as a manipulation check. State cognitive dissonance was measured with items framed around respondents' immediate feelings about the trade-offs embedded in the purchase scenarios. Health literacy was assessed using a Chinese-adapted Newest Vital Sign instrument. These measures were used to contextualize segment differences and mechanism interpretation rather than to substitute for revealed preference evidence from the DCE.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e2.6. Econometric Modeling and Statistical Analysis\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eChoices were analyzed under RUT using a Random Parameters Logit (RPL) to accommodate preference heterogeneity and relax the IIA assumption. Utility for individual \u003cem\u003en\u003c/em\u003e, alternative \u003cem\u003ej\u003c/em\u003e, and choice task \u003cem\u003et\u003c/em\u003e was specified as Eq.\u0026nbsp;(\u003cspan refid=\"Equ1\" class=\"InternalRef\"\u003e1\u003c/span\u003e),\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Equ1\" class=\"Equation\"\u003e \u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$\\:{U}_{njt}\\:=\\:{V}_{njt}\\:+\\:{\\epsilon\\:}_{njt}$$\u003c/div\u003e \u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003ewith the systematic component modeled as a function of price, Nutri-Grade, sweetener strategy, flavor, and their interactions. Specifically, we estimated \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{V}_{njt}\\)\u003c/span\u003e\u003c/span\u003e through Eq.\u0026nbsp;(\u003cspan refid=\"Equ2\" class=\"InternalRef\"\u003e2\u003c/span\u003e),\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Equ2\" class=\"Equation\"\u003e \u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e\n$$\\:{V}_{njt}=\\alpha\\:{Price}_{njt}+{\\beta\\:}_{1}{Grade}_{njt}+{\\beta\\:}_{2}{Sweetener}_{njt}+{\\beta\\:}_{3}({Grade}_{njt}\\times\\:{Sweetener}_{njt})+\\gamma\\:{Flavor}_{njt}+{ASC}_{opt-out}$$\u003c/div\u003e \u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{ASC}_{opt-out}\\)\u003c/span\u003e\u003c/span\u003e captures baseline propensity to reject both products. The key test of the clean-label paradox is the interaction between favorable grades\u0026mdash;particularly Grade A\u0026mdash;and the artificial sweetener blend (aspartame\u0026thinsp;+\u0026thinsp;acesulfame K). A significantly negative interaction term indicates that the positive utility of a high Nutri-Grade is systematically discounted when paired with additive-salient ingredients, consistent with the proposed dissonance-based mechanism\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e. Marginal willingness-to-pay (WTP) for attribute levels was calculated as Eq.\u0026nbsp;(\u003cspan refid=\"Equ3\" class=\"InternalRef\"\u003e3\u003c/span\u003e),\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Equ3\" class=\"Equation\"\u003e \u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e\n$$\\:{WTP}_{k}\\:=\\:{-\\beta\\:}_{k}\\:/\\:\\alpha\\:$$\u003c/div\u003e \u003cdiv class=\"EquationNumber\"\u003e3\u003c/div\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003ewith uncertainty quantified using the delta method or bootstrapping. To complement continuous heterogeneity in the RPL, we also estimated a Latent Class Model (LCM) to identify distinct preference segments, such as \"Nutri-Grade followers\" versus \"clean-label purists,\" with the number of classes chosen based on information criteria and interpretability; segment profiles were then compared on consumption intensity and the post-choice psychological measures\u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e. All analyzes were conducted in Python (PyBiogeme) and Stata 18.0, and robustness checks included alternative coding schemes (effects versus dummy coding), re-estimation excluding respondents flagged by quality screens, and comparisons across MNL, RPL, and LCM specifications.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e2.7. Ethics\u003c/h2\u003e \u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e Ethical review and approval were waived by the Institutional Review Board of Tianjin University of Traditional Chinese Medicine due to the anonymous and minimal-risk nature of the study. All participants provided electronic informed consent prior to participation. All methods were carried out in accordance with the relevant guidelines and regulations, and in accordance with the Declaration of Helsinki.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Sample characteristics and data quality\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eA total of 3,000 respondents completed the survey. After applying the pre-specified quality-control protocol (speeding, failed instructed-response attention checks, and patterned responding in psychometric batteries), 264 responses were excluded and the final analytic sample consisted of 2,736 respondents (retention rate 91.2%). The sample achieved a near-even gender split (49.1% male; 50.9% female) and broad geographic coverage across city tiers (Tier 1 to Tier 3 and below). As summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, respondents were relatively engaged with the product category: 41.1% reported heavy CSD consumption (\u0026ge;\u0026thinsp;3 times/week), and 61.3% reported checking nutrition labels often or always. The age distribution aligned with quota targets (mean age 29.4 years, SD 6.8), and educational attainment skewed toward higher education, consistent with patterns commonly observed in opt-in digital panels (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSample characteristics and category engagement.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFrequency (n)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePercentage (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eA. Socio-Demographics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,344\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e49.12%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,392\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e50.88%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge Group\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18\u0026ndash;24 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e887\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e32.42%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e25\u0026ndash;34 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,113\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e40.68%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e35\u0026ndash;45 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e736\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e26.90%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarital Status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSingle / Unmarried\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,434\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e52.41%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMarried / Partnered\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,225\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e44.77%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDivorced / Other\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.81%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCity Tier\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTier 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e608\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e22.22%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNew Tier 1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e724\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e26.46%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTier 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e775\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e28.33%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTier 3 and below\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e629\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e22.99%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation Level\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHigh school or below\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e312\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e11.40%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAssociate degree / College\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e588\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e21.49%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBachelor's degree\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,521\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e55.59%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMaster's degree or higher\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e315\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e11.51%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMonthly Household Income\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;10,000 RMB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e843\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e30.81%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10,000\u0026ndash;19,999 RMB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,082\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e39.55%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;20,000 RMB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e29.64%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eB. Health Status \u0026amp; Lifestyle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI Status \u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUnderweight (18.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e251\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e9.17%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNormal weight (18.5\u0026ndash;23.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,678\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e61.33%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverweight/Obese (\u0026ge;\u0026thinsp;24.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e807\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e29.50%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFamily History of Diabetes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e755\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e27.60%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo / Unsure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,981\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e72.40%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCurrent Dieting Status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOn a diet\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e980\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e35.82%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNot currently dieting\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,756\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e64.18%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSelf-Rated Health\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eExcellent / Very Good\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e866\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e31.65%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGood\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,352\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e49.42%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFair / Poor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e518\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e18.93%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eC. CSD Category Engagement\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConsumption Frequency\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLight (1\u0026ndash;3 times/month)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e788\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e28.80%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModerate (1\u0026ndash;2 times/week)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e823\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e30.08%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHeavy (\u0026ge;\u0026thinsp;3 times/week)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,125\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e41.12%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrimary Purchase Channel\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eConvenience Store (C-Store)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,768\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e64.62%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSupermarket / Hypermarket\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e553\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e20.21%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE-commerce / O2O Delivery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e415\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e15.17%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLabel Reading Habit\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNever / Rarely\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e416\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e15.20%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSometimes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e643\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e23.50%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOften / Always\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1,677\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e61.29%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eNote: Percentages may not sum to 100 due to rounding. City tier follows the Tier 1 / New Tier 1 / Tier 2 / Tier 3 and below classification used in quota sampling. \"Heavy consumption\" denotes carbonated soft drink (CSD) intake\u0026thinsp;\u0026ge;\u0026thinsp;3 times/week. BMI categories follow the Working Group on Obesity in China (WGOC) guideline cut-points.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Measurement properties and manipulation checks\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eBefore estimating choice models, we verified that post-choice constructs and experimental manipulations supported the intended interpretation of \"ingredient naturalness cues\" and mechanism-related measures. Confirmatory factor analysis indicated that the three-factor measurement model (State Cognitive Dissonance, Perceived Naturalness, and Food Tech Neophobia) fit the data well, with standard indices meeting conventional benchmarks; internal consistency and convergent validity were also strong, with Cronbach's α, composite reliability, and AVE values exceeding commonly used thresholds (see Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePsychometric properties of post-choice measures.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLatent Construct and Items\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStd. Loading (λ)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCronbach's α\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCR\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAVE\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eState Cognitive Dissonance (SCD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.893\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.893\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.732\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSCD1: I felt conflicted choosing this drink because of its ingredients.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.841\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSCD2: I was uncomfortable with the trade-off between grade and additives.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.882\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSCD3: The choice left me feeling uneasy about the product's naturalness.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.853\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerceived Naturalness (PN)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.921\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.921\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePN1: This product feels natural to me.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.916\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePN2: I perceive this product as free from artificial processing.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.883\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePN3: This product aligns with my idea of \"clean label\".\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.878\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFood Tech Neophobia (FTN)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.865\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.872\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.685\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFTN1: Artificial sweeteners in food are unnecessary.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.819\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFTN2: Society is too dependent on food technology.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.836\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFTN3: New food additives are something I try to avoid.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.843\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNotes: Std. Loading\u0026thinsp;=\u0026thinsp;standardized factor loading from CFA; α\u0026thinsp;=\u0026thinsp;Cronbach's alpha; CR\u0026thinsp;=\u0026thinsp;composite reliability; AVE\u0026thinsp;=\u0026thinsp;average variance extracted. All loadings are statistically significant (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eCrucially, the sweetener attribute generated a pronounced and theoretically consistent naturalness gradient. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, cane sugar was perceived as the most natural formulation (M\u0026thinsp;=\u0026thinsp;5.82), followed by the stevia/erythritol blend (M\u0026thinsp;=\u0026thinsp;5.43), with HFCS rated intermediate (M\u0026thinsp;=\u0026thinsp;3.64). The artificial sweetener blend (aspartame\u0026thinsp;+\u0026thinsp;acesulfame K) was rated substantially less natural than all other conditions (M\u0026thinsp;=\u0026thinsp;2.15), and Bonferroni-adjusted comparisons indicated that it differed significantly from every other sweetener strategy (all p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This manipulation check confirms that the DCE operationalized ingredient-based \"naturalness versus artificiality\" cues in a way that is perceptually salient to respondents, establishing the necessary condition for testing whether nutrient-grade effects are contingent on reformulation strategy.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Baseline preference structure: main-effects random parameters logit\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eWe first estimated a random parameters logit (RPL) model with main effects to characterize baseline preference patterns. Full estimates are reported in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The price coefficient was negative and statistically significant (β = \u0026minus;0.184, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating economically coherent choice behavior within the observed price range. The opt-out constant was large and negative (ASC\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;2.56, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), consistent with a realistic tendency to reject both offered products in some tasks rather than being forced into a purchase.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline preferences for nutrient grades and sweetener strategies: random parameters logit.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAttribute\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLevel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCoef. (β)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS.E.\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ez-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e95% Conf. Interval\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrice (RMB)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.184\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.042\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-4.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.266, -0.102]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNutri-Grade\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGrade A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.315\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.058\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[ 0.201, 0.429]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e(Ref: Grade C)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGrade B\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.072\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.145\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.025, 0.169]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGrade D\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.248\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.056\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-4.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.358, -0.138]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNone (No Label)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.062\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-0.39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.698\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.146, 0.098]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSweetener\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCane Sugar\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.295\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.052\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[ 0.193, 0.397]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e(Ref: HFCS)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNatural Blend (Stevia)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.045\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.049\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.362\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.051, 0.141]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eArtificial Blend (Aspartame)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.308\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.065\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-4.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.435, -0.181]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFlavor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCitrus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.041\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.714\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.065, 0.095]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e(Ref: Classic)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePeach\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.068\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.089\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-0.010, 0.146]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConstant\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOpt-out (ASC)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-2.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-14.22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[-2.913, -2.207]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom Parameters (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSD_Price\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.025\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.405\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSD_Grade A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.485\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.092\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSD_Sweetener (Artificial)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.115\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eModel Fit Statistics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eObservations\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c7\" namest=\"c3\"\u003e \u003cp\u003e32,832\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eLog-Likelihood\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c7\" namest=\"c3\"\u003e \u003cp\u003e-26,104.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eMcFadden Pseudo R\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c7\" namest=\"c3\"\u003e \u003cp\u003e0.168\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eAIC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c7\" namest=\"c3\"\u003e \u003cp\u003e52,245.00\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003eBIC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c7\" namest=\"c3\"\u003e \u003cp\u003e52,396.70\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"7\"\u003eNotes: The dependent variable is the choice among two product alternatives plus an opt-out option (\"buy neither\"). Reference levels: Nutri-Grade\u0026thinsp;=\u0026thinsp;Grade C; Sweetener strategy\u0026thinsp;=\u0026thinsp;HFCS; Flavor\u0026thinsp;=\u0026thinsp;Classic. The price is in RMB. The opt-out term is an alternative-specific constant (ASC). Random-parameter standard deviations (SD) are reported for parameters specified as random. Bold indicates p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eNutri-Grade effects exhibited a non-linear response pattern rather than a smooth monotonic gradient. Relative to Grade C (reference), Grade A generated a robust positive utility premium (β\u0026thinsp;=\u0026thinsp;0.315, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while Grade B was small and not statistically distinguishable from zero (β\u0026thinsp;=\u0026thinsp;0.072, p\u0026thinsp;=\u0026thinsp;0.145). In contrast, Grade D was significantly penalized (β = \u0026minus;0.248, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and the \"no label\" condition was not significantly different from Grade C. This \"threshold-like\" pattern suggests that, in this category, consumers place disproportionate value on the top grade while treating intermediate grades as less salient.\u003c/p\u003e \u003cp\u003eSweetener strategy contributed independently to utility. Cane sugar was preferred over HFCS (β\u0026thinsp;=\u0026thinsp;0.295, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), whereas the artificial sweetener blend carried a significant disutility (β = \u0026minus;0.308, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The stevia/erythritol blend did not yield a statistically significant improvement relative to HFCS in the main-effects model (β\u0026thinsp;=\u0026thinsp;0.045, p\u0026thinsp;=\u0026thinsp;0.362). Importantly, the random-parameter estimates indicated substantial unobserved heterogeneity for Grade A and for the artificial sweetener blend, consistent with polarized responses toward artificial sweeteners. Flavor coefficients were comparatively small, implying that choices were driven primarily by the health\u0026ndash;ingredient\u0026ndash;price bundle rather than flavor variation.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.4. Conditional effectiveness: grade \u0026times; sweetener interactions (\"clean-label paradox\")\u003c/h2\u003e \u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eTo test whether the effect of nutrient grades depends on reformulation cues, we extended the RPL specification by including interactions between nutrient grade levels and sweetener strategies. Adding these interaction terms substantially improved model fit relative to the main-effects specification (likelihood-ratio test χ\u0026sup2;(6)\u0026thinsp;=\u0026thinsp;142.5, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating that consumers do not evaluate nutrient grades independently of ingredient composition. Key interaction estimates are reported in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eConditional effectiveness of nutrient grades: key grade \u0026times; sweetener interaction effects in the RPL model.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParameter\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCoef. (β)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS.E.\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ez-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMain Effects (Baseline)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNutri-Grade A\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.425\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.062\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNutri-Grade B\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.084\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.055\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.126\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArtificial Blend (Main)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.215\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-3.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e0.002\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInteraction Effects (The Paradox)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade A \u0026times; Artificial Blend\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.485\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.088\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-5.51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade B \u0026times; Artificial Blend\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.185\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.082\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-2.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e0.024\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade A \u0026times; Natural Blend (Stevia)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.052\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.114\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.648\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade B \u0026times; Natural Blend (Stevia)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.108\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.889\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel Fit\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLog-Likelihood\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c5\" namest=\"c2\"\u003e \u003cp\u003e-25,985.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePseudo R\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c5\" namest=\"c2\"\u003e \u003cp\u003e0.172\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNotes: Estimates are from the interaction specification that adds cross-attribute terms between Nutri-Grade levels and sweetener strategies. The focal \"clean-label paradox\" test is the interaction between favorable grades (especially Grade A) and the artificial sweetener blend (aspartame\u0026thinsp;+\u0026thinsp;acesulfame K); a negative coefficient indicates discounting of the grade premium under ingredient conflict. The model controls for price, flavor, and main effects; only theoretically central interactions are displayed. Bold indicates p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe central result is a strong negative interaction between Grade A and the artificial sweetener blend (β = \u0026minus;0.485, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This coefficient implies that the utility premium attached to the top nutrient grade is materially discounted when the product contains additive-salient artificial sweeteners. In practical terms, the Grade A signal does not \"stack\" additively on top of an artificial sweetener formulation: the interaction penalty offsets much of the Grade A main effect, producing only a minimal net gain from upgrading the grade under the artificial blend. By contrast, interactions between favorable grades and the stevia/erythritol blend were small and not statistically significant, suggesting that the discounting is not an inherent response to sugar reduction per se but is concentrated in the presence of artificial additives.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThis conditional pattern is also reflected in predicted choice probabilities. As illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, improvements in Nutri-Grade from C to A are associated with a clear rise in predicted choice probability for ingredient-congruent formulations (e.g., cane sugar), whereas the corresponding trajectory for the artificial sweetener blend is comparatively flat. This visualization reinforces the interpretation of nutrient grades as conditional signals: the demand response to a favorable grade is substantially weaker when ingredient cues conflict with the algorithmic health rating.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e3.5. Willingness-to-pay: quantifying the valuation erosion\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eTo translate preference parameters into economically interpretable magnitudes, we computed marginal willingness-to-pay (WTP) for key contrasts using the ratio of attribute coefficients to the (negative) price coefficient. Results are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e and visualized in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eWillingness-to-pay contrasts quantify valuation erosion when Grade A is achieved via artificial sweeteners.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eScenarios \u0026amp; Contrasts\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEstimated WTP (RMB)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% Conf. Interval\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1. The \"Ideal\" Health Premium\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade A vs. Grade C (with Cane Sugar)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;2.31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[1.85,2.77]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2. The \"Paradox\" Scenario\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade A vs. Grade C (with Artificial Blend)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.27 \u003csup\u003ens\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;0.15,0.69]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3. The \"Paradox Penalty\" (Diff)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e(Scenario 1) minus (Scenario 2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-2.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;2.55,\u0026minus;1.53]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4. Reformulation Value\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNatural Blend (Stevia) vs. Artificial Blend\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;1.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[1.22,2.14]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003eNote: Estimates based on the ratio of coefficients from the RPL interaction model. \u003csup\u003ens\u003c/sup\u003e indicates the WTP estimate is not significantly different from zero at \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eIn an ingredient-congruent context (Grade A relative to Grade C with cane sugar), respondents exhibited a sizable premium of +\u0026thinsp;2.31 RMB (95% CI [1.85, 2.77]), indicating that the top grade carries meaningful economic value when it aligns with expected ingredient cues. However, when the same Grade A upgrade occurs under the artificial sweetener blend, the implied premium collapses to +\u0026thinsp;0.27 RMB and is not statistically distinguishable from zero (95% CI [\u0026minus;\u0026thinsp;0.15, 0.69]). The difference between these scenarios implies a valuation erosion of approximately \u0026minus;\u0026thinsp;2.04 RMB (95% CI [\u0026minus;\u0026thinsp;2.55, \u0026minus;\u0026thinsp;1.53]), which quantifies the economic magnitude of the grade\u0026ndash;ingredient conflict. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents these WTP contrasts directly, highlighting the sharp divergence in the monetary value of Grade A across reformulation contexts.\u003c/p\u003e \u003cp\u003eWTP estimates also inform the valuation of alternative reformulation pathways. Replacing the artificial sweetener blend with the stevia/erythritol blend yields a positive WTP uplift of +\u0026thinsp;1.68 RMB (95% CI [1.22, 2.14]), consistent with a meaningful consumer premium for sugar-reduction strategies that are more compatible with clean-label interpretations.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e3.6. Heterogeneity: latent class segmentation and interpretation\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eGiven the strong heterogeneity suggested by the random-parameters estimates, we estimated latent class models to identify discrete segments with qualitatively different evaluation rules. Model comparison favored a three-class solution balancing fit and interpretability. Segment sizes and defining preference parameters are reported in Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e, with psychometric profiling summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eLatent class segmentation reveals heterogeneous decision rules and identifies the segment driving grade discounting.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAttributes\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eClass 1:\u003c/p\u003e \u003cp\u003eClean-Label Purists\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClass 2:\u003c/p\u003e \u003cp\u003eAlgorithmic Health Seekers\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eClass 3:\u003c/p\u003e \u003cp\u003ePrice-Sensitive Traditionalists\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSegment Size (Share)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e44.20%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e32.50%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23.30%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKey Preferences (β)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrice\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.65 ***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNutri-Grade A (Main)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.45 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.75 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e+\u0026thinsp;0.18 *\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCane Sugar\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.82 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e+\u0026thinsp;0.55 ***\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArtificial Blend (Main)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.85 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.15 \u003csup\u003ens\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.25 *\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThe Paradox Interaction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGrade A \u0026times; Artificial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.92 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.08 \u003csup\u003ens\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.12 \u003csup\u003ens\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInterpretation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\"A-Grade is fake if artificial\"\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\"I trust the A-Grade\"\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\"Too expensive / I want real sugar\"\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003e*** \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001, * \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, \u003csup\u003ens\u003c/sup\u003e not significant. Bold values indicate the defining characteristic of each class.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe largest segment (\"Clean-Label Purists\", 44.2%) exhibited a pronounced aversion to the artificial sweetener blend (β = \u0026minus;1.85, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and the strongest discounting of Grade A when paired with artificial sweeteners (Grade A \u0026times; artificial β = \u0026minus;0.92, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Consistent with the proposed cue-conflict mechanism, this segment also displayed higher post-choice state cognitive dissonance. A second segment (\"Algorithmic Health Seekers\", 32.5%) placed strong positive value on Grade A (β\u0026thinsp;=\u0026thinsp;0.75, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) while showing little evidence of interaction-based discounting and relatively weak sensitivity to artificial sweeteners, consistent with reliance on the front-of-pack grade as a primary heuristic. A third segment (\"Price-Sensitive Traditionalists\", 23.3%) was characterized by the strongest price sensitivity and a preference for cane sugar formulations, with comparatively muted responsiveness to nutrient grades. Together, these segments indicate that the clean-label paradox is not uniform across the population: it is driven primarily by a large segment that treats artificial sweeteners as a salient negative ingredient cue and correspondingly discounts favorable algorithmic grades.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e3.7. Robustness checks\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eRobustness checks indicate that the central inference\u0026mdash;the negative Grade A \u0026times; artificial sweetener interaction\u0026mdash;does not depend on a single modeling choice. As shown in Appendix Table A1, the key interaction remains negative and statistically significant across MNL, RPL, and latent class specifications. Sensitivity analyses using effects coding yield substantively similar conclusions (Appendix Table A2), and re-estimation on stricter quality-screened subsamples produces closely aligned effect sizes and WTP penalties (Appendix Table A3).\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e4.1. Main findings and why they matter for front-of-pack policy evaluation\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThis study provides evidence that front-of-pack (FOP) nutrient grades in beverages operate as conditional signals rather than unconditional demand-side cues. In a discrete choice setting, respondents valued a favorable nutrient grade (especially Grade A), but the magnitude of that valuation depended critically on how the grade was achieved. When the grade improvement was paired with an artificial sweetener blend, the Grade A premium was sharply attenuated (Grade A \u0026times; artificial blend interaction β = \u0026minus;0.485, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and the associated willingness-to-pay premium for upgrading from Grade C to Grade A effectively collapsed\u0026mdash;from +\u0026thinsp;2.31 RMB under cane sugar to +\u0026thinsp;0.27 RMB (not statistically distinguishable from zero) under artificial sweeteners, implying a valuation erosion of approximately 2.04 RMB.\u003c/p\u003e \u003cp\u003eFrom a Food Policy perspective, the key implication is not simply that \"consumers dislike artificial sweeteners.\" Rather, the results speak to an instrument-design issue: the effectiveness of nutrient profiling policies depends on the coherence between the algorithmic signal that policy puts on the front of pack and the ingredient cues that consumers treat as relevant to healthfulness, legitimacy, or trust\u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e. If a scoring system is implemented in a policy environment where manufacturers can most readily secure favorable grades via additive-salient reformulation pathways, then demand-side responses to the policy may be weaker than expected even when the policy is implemented correctly and information is technically accurate\u003csup\u003e[\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e4.2. Mechanism interpretation: credibility conflict as a behavioral constraint on policy impact\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe pattern observed here is consistent with a credibility-conflict mechanism: consumers appear to discount the informational value of a favorable grade when it is paired with ingredient cues interpreted as \"artificial.\" The manipulation check confirms that the artificial blend is perceived as substantially less natural than alternative sweetening strategies, providing a credible basis for the cue-conflict interpretation\u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThis matters because most policy evaluations of nutrient profiling labels implicitly assume separability: a better grade should increase perceived healthfulness and choice probability regardless of how the product was reformulated. Our results challenge that separability assumption in a concrete way. The negative interaction indicates that the grade premium does not transfer across reformulation contexts; instead, the grade's marginal value is contingent on ingredient congruence\u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e. In practice, this implies that nutrient grades can lose part of their \"behavioral purchasing power\" precisely in the situations where policy expects them to matter\u0026mdash;when products are reformulated to obtain better scores.\u003c/p\u003e \u003cp\u003eNotably, the attenuation is not a generic response to sugar reduction. Interactions involving the stevia/erythritol blend were not significant, suggesting that consumers do not automatically penalize \"lower sugar\" reformulation\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e. The discounting appears concentrated in reformulation pathways that activate salient beliefs about additives and artificial processing.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.3. Reformulation pathways as a missing link in label policy effectiveness\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eA central contribution of this study is to treat reformulation pathway choice as part of the policy mechanism rather than as an external industry detail. Nutrient-grade policies do not only inform consumers; they also shape manufacturer incentives by rewarding particular compositional changes. In beverages, the least costly and most technically straightforward pathway to improved nutrient scores is often substitution with non-nutritive sweeteners\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e. Our findings indicate that this pathway may generate a demand-side penalty that erodes the benefit of achieving a favorable grade.\u003c/p\u003e \u003cp\u003eThis suggests an important refinement to how nutrient profiling policies are conceptualized and evaluated: implementation success should not be assessed solely by uptake of labels or by the proportion of products achieving better grades. Instead, evaluation should also track the distribution of reformulation pathways used to obtain those grades and whether the demand response differs across those pathways\u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e. If a policy disproportionately induces \"score-improving\" reformulation that is viewed skeptically by a large consumer segment, aggregate improvements in on-pack grades may overstate real-world shifts in purchasing and welfare.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e4.4. Heterogeneity and distributional consequences: who follows the grade, and who contests it\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eAverage effects can obscure meaningful distributional patterns. The latent class results highlight three segments with different decision rules, and these segments are policy-relevant. A large segment of \"clean-label purists\" (44.2%) drives the paradox: they strongly dislike the artificial sweetener blend and exhibit the strongest within-segment discounting of Grade A when paired with artificial sweeteners. In contrast, \"algorithmic health seekers\" (32.5%) respond strongly to Grade A and do not show meaningful discounting, while \"price-sensitive traditionalists\" (23.3%) are driven primarily by price and prefer cane sugar with comparatively muted responsiveness to nutrient grades.\u003c/p\u003e \u003cp\u003eThis heterogeneity suggests that nutrient grading policies may produce uneven outcomes across groups even when the label is equally visible. If a sizable share of consumers systematically contests or discounts grades for certain reformulation strategies, then policy effects may be concentrated among those who treat the grade as a trusted heuristic, while other consumers remain anchored to ingredient-based evaluation\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]\u003c/sup\u003e. This has practical implications for both equity and effectiveness: a \"one-size-fits-all\" informational tool may not generate uniform behavioral responses, and market adjustments (e.g., product offerings and pricing) may interact with these segment differences.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e4.5. Policy implications: designing and evaluating nutrient grades under reformulation and contestation\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe findings point to a broader lesson for food labeling policy: front-of-pack nutrient grades are not merely informational devices; they are policy instruments that jointly shape consumer inference and firm behavior. In that sense, the key design problem is not \"whether a grade is accurate,\" but whether the grade remains credible and behaviorally operative when firms pursue the most cost-effective pathways to improve it\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/sup\u003e. Our evidence indicates that the demand-side value of a top grade can be substantially discounted when the grade is produced through additive-salient reformulation (artificial sweetener substitution), with the economic signature of this discounting visible in the collapse of the Grade A premium in WTP terms.\u003c/p\u003e \u003cp\u003eA useful way to translate this into policy design is to view nutrient grades as signals with endogenous production. The \"signal\" (Grade A) is generated by an algorithm, but the conditions under which products achieve the signal are shaped by industry incentives and constraints (cost, formulation feasibility, taste preservation). When the cheapest route to a stronger signal produces ingredient cues that many consumers interpret as inconsistent with \"healthfulness,\" the policy can encounter a credibility constraint: the signal is present, yet its marginal persuasive power is reduced. This implies that the real-world effect of a grading system depends on (i) the scoring rule and threshold structure, (ii) the distribution of feasible reformulation options, and (iii) prevailing consumer beliefs about the legitimacy of those options.\u003c/p\u003e \u003cp\u003eThree implications follow for policy design and implementation.\u003c/p\u003e \u003cp\u003eFirst, reformulation pathways should be treated as a core outcome in policy evaluation, not a background condition. Many evaluations implicitly equate improved grades on shelves with improved public health impact. Our results caution against that equivalence: if grade improvements are disproportionately achieved through reformulation pathways that trigger skepticism, the policy may produce \"paper gains\" in grades with limited shifts in demand. A practical evaluation framework should therefore report, alongside grade distributions, the composition of reformulation strategies used to reach those grades (e.g., sugar reduction via sweeteners versus other adjustments) and test whether demand responses differ across those pathways.\u003c/p\u003e \u003cp\u003eSecond, the policy architecture may need complementary interpretive scaffolding to preserve the grade's meaning under cue conflict. The underlying issue is not that consumers are \"irrational,\" but that they integrate multiple cues. If the grade is interpreted as a holistic endorsement while ingredient cues trigger \"artificiality\" concerns, conflict is predictable. Policy can respond by improving interpretability of what the grade represents (a nutrient-profile summary) and, importantly, what it does not represent. In practice, this does not require turning the front-of-pack into a complex label. It can involve standardized explanatory language in policy rollouts, consistent public communication, and\u0026mdash;where appropriate\u0026mdash;harmonization with other informational schemes that address processing-related concerns (so consumers are not forced to infer \"processing\" from ingredients while inferring \"health\" from grades).\u003c/p\u003e \u003cp\u003eThird, distributional heterogeneity implies that a single-grade instrument can yield uneven behavioral gains and potentially uneven welfare effects. The latent classes indicate that a large segment (\"clean-label purists\") drives the discounting of favorable grades when paired with artificial sweeteners, whereas another segment appears to follow the grade as a primary heuristic. Policy evaluation that reports only average effects may therefore mischaracterize the mechanism: the policy works strongly for some consumers but is contested by others, particularly where reformulation relies on additives. This has two practical implications. On the demand side, it suggests that complementary communication may need to be targeted or tailored. On the supply side, it suggests that manufacturers' optimal responses to a grading policy may involve product-line differentiation (e.g., parallel offerings that appeal to grade-followers versus ingredient-first consumers), which can reshape market structure and pricing. Policymakers should anticipate such endogenous adjustment rather than assuming a uniform shift.\u003c/p\u003e \u003cp\u003eTaken together, these implications emphasize that the effectiveness of nutrient grading policies depends on the interaction between algorithm design and industry response. The central policy takeaway is not that grades \"fail,\" but that they can become conditional and contested when score-improving reformulation is achieved via ingredients that consumers treat as qualitatively meaningful.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e4.6. Limitations and scope conditions\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eSeveral limitations delineate the scope conditions of our inferences and point to where additional evidence is needed.\u003c/p\u003e \u003cp\u003eFirst, discrete choice experiments approximate market decision-making but cannot fully capture dynamic purchasing and learning. Real consumers may update beliefs over repeated exposure (e.g., learning that a \"Grade A\u0026thinsp;+\u0026thinsp;artificial sweeteners\" product tastes good, or revising beliefs about sweeteners after information shocks). Our design captures preference trade-offs at the point of choice under controlled information, which is appropriate for isolating conditional signaling, but it does not model dynamic belief formation.\u003c/p\u003e \u003cp\u003eSecond, we intentionally used generic products to isolate policy-relevant cues; this strengthens internal validity but limits the ability to capture brand trust and reputation as moderators. In real markets, brand equity can either buffer skepticism (trusted brands may \"carry\" a grade further) or intensify it (brands associated with \"naturalness\" may face stronger backlash when using artificial additives). Whether brand moderates the grade\u0026ndash;ingredient interaction is therefore an open empirical question.\u003c/p\u003e \u003cp\u003eThird, external validity depends on the information environment. Perceptions of artificial sweeteners and \"clean label\" norms are shaped by media discourse, regulatory debates, and health communication. The magnitude of discounting is likely to be state-dependent: periods of heightened sweetener controversy or misinformation may amplify skepticism, whereas credible communication or shifting norms could attenuate it. Our results identify a robust behavioral pattern under the measured perceptions in our sample; translating effect sizes across contexts requires attention to those perception baselines.\u003c/p\u003e \u003cp\u003eFourth, the policy mechanism is inherently general-equilibrium: grading policies shape firm incentives, which can alter product portfolios, ingredient choices, and pricing. Our DCE holds many market features fixed to identify preference parameters, but in reality firms may respond strategically\u0026mdash;e.g., by adjusting prices to offset demand penalties, changing package claims to reframe ingredient cues, or reallocating marketing budgets. These strategic responses could either mitigate or exacerbate the conditional effectiveness documented here. Recognizing this is important: the conditional signal is not just a consumer cognition issue; it is embedded in a market system where firms adapt.\u003c/p\u003e \u003cp\u003eFinally, while the post-choice measures support a cue-conflict interpretation, they remain correlational with respect to mechanism. A more stringent mechanism test would experimentally manipulate interpretive frames (e.g., providing standardized explanations of what the grade reflects, or clarifying the safety/regulatory status of sweeteners) to test whether discounting is reduced without sacrificing label simplicity.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e4.7. Future research\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThe results motivate a research agenda that connects micro-level conditional signaling to macro-level policy performance.\u003c/p\u003e \u003cp\u003eField validation and repeated-choice behavior. A natural next step is to test whether the grade\u0026ndash;ingredient interaction predicts realized purchasing in settings with real stakes\u0026mdash;online grocery experiments, controlled shopping tasks with incentives, or retail scanner/panel data where available. Such studies could examine not only one-shot choice but also repeat purchase, substitution patterns across beverage categories, and persistence over time.\u003c/p\u003e \u003cp\u003eInterpretability interventions. The credibility-conflict mechanism suggests that the effectiveness of nutrient grades may be improved by interventions that reduce misalignment between what the grade communicates and what consumers infer from ingredients. Future experiments could randomize brief, standardized interpretive statements (e.g., clarifying that the grade summarizes nutrients rather than processing), or compare alternative front-of-pack designs that better anticipate contested attributes. Importantly, the goal is to test whether interpretability can be increased without increasing cognitive load\u0026mdash;an essential constraint for policy tools designed for rapid decision contexts.\u003c/p\u003e \u003cp\u003eEndogenous reformulation and policy simulation. The most policy-relevant extension is to integrate demand estimates with a stylized supply response model. Because grading policies are likely to change the payoff to specific reformulation pathways, a complete evaluation should ideally combine (i) consumer demand response conditional on reformulation and (ii) firms' cost-minimizing reformulation choices. Even a simplified simulation\u0026mdash;where firms choose between sugar reduction pathways with different cost and demand consequences\u0026mdash;could illuminate when a grading policy is likely to induce \"behaviorally robust\" reformulation versus \"behaviorally fragile\" score improvements. This would move the literature closer to actionable ex ante policy assessment.\u003c/p\u003e \u003cp\u003eCross-category generalization. Finally, the conditional signal framework can be tested beyond beverages in categories where reformulation involves contested additives or processing cues (e.g., fat replacers, emulsifiers, preservatives). If conditional effectiveness is common, it strengthens the case for policy evaluation frameworks that explicitly treat algorithmic labels as signals whose credibility depends on how products achieve them.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eThis study shows that front-of-pack nutrient grades in carbonated soft drinks function as conditional signals: favorable grades are valued, but their marginal influence is substantially reduced when achieved through additive-salient artificial sweetener reformulation. The resulting collapse of the Grade A WTP premium under artificial sweeteners quantifies the economic significance of this credibility conflict, while latent class results reveal that discounting is driven by a large consumer segment.\u003c/p\u003e \u003cp\u003eFor policy, the central message is that nutrient grading systems cannot be evaluated solely on label uptake or grade distributions. Their realized impact depends on the interaction between scoring rules and the reformulation pathways they induce, as well as on how consumers integrate algorithmic ratings with ingredient cues. Designing and evaluating nutrient grading policies with these behavioral and market-contingent mechanisms in mind is likely to be necessary for translating improved algorithmic scores into sustained shifts in purchasing and meaningful public health gains.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eThe following abbreviations are used in this manuscript:\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"524\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eAceK\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eAcesulfame K (acesulfame potassium)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eAIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eAkaike Information Criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eANOVA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eAnalysis of Variance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eASC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eAlternative-Specific Constant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eAVE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eAverage Variance Extracted\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eAvePP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eAverage Posterior Probability\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eBayesian Information Criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eBody Mass Index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eC-Store\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eConvenience Store\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eCFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eConfirmatory Factor Analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eCFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eComparative Fit Index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eConfidence Interval\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eCR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eComposite Reliability\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eCSD(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eCarbonated Soft Drink(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eDCE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eDiscrete Choice Experiment\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eFOPL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eFront-of-Pack Labeling\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eFTN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eFood Technology Neophobia\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eHFCS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eHigh Fructose Corn Syrup\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eIIA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eIndependence of Irrelevant Alternatives\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eIRB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eInstitutional Review Board\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eLCM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eLatent Class Model\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eLR test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eLikelihood Ratio test\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eMNL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eMultinomial Logit model\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eMRS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eMarginal Rate of Substitution\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eNVS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eNewest Vital Sign\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eO2O\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eOnline-to-Offline\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003ePN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003ePerceived Naturalness\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eRMSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eRoot Mean Square Error of Approximation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eRPL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eRandom Parameters Logit (Mixed Logit)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eRUT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eRandom Utility Theory\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eSCD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eState Cognitive Dissonance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eStandard Deviation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eStandard Error\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eSRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eStandardized Root Mean Square Residual\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eTLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eTucker\u0026ndash;Lewis Index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eWGOC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eWorking Group on Obesity in China\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11.8547%;\"\u003e\n \u003cp\u003eWTP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88.1453%;\"\u003e\n \u003cp\u003eWillingness to Pay\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003e \u003cb\u003eInstitutional Review Board Statement\u003c/b\u003e \u003c/h2\u003e \u003cp\u003e Ethical review and approval were waived for this study by the Institutional Review Board of Tianjin University of Traditional Chinese Medicine due to the anonymous nature of the data collection and the minimal risk posed to participants.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eInformed Consent\u003c/h2\u003e \u003cp\u003e \u003cb\u003eStatement\u003c/b\u003e: Informed consent was obtained from all subjects involved in the study.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eConflicts of Interest:\u003c/h2\u003e \u003cp\u003eThe authors declare no conflicts of interest.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding:\u003c/h2\u003e \u003cp\u003eThis research received no external funding.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConceptualization, Y.C. and M.W.; methodology, Y.C. and T.Z.; software, Y.C. and H.G.; validation, H.G., Z.L. and K.M.; formal analysis, Y.C.; investigation, Y.C., H.G. and Z.L.; resources, T.Z. and M.W.; data curation, H.G. and Z.L.; writing\u0026mdash;original draft preparation, Y.C.; writing\u0026mdash;review and editing, K.M., M.W. and T.Z.; visualization, Y.C. and Z.L.; supervision, T.Z. and M.W.; project administration, T.Z. and M.W.; funding acquisition, T.Z. and M.W. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgments:\u003c/h2\u003e \u003cp\u003eWe would like to express our gratitude to the operational team at Wenjuanxing for their technical support in data collection and panel management. We also thank all the anonymous participants who took part in this survey. During the preparation of this manuscript, the authors used Gemini 3 for the purpose of English language editing and proofreading. The authors have reviewed and edited the output and take full responsibility for the content of this publication.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical considerations regarding the participants.\u003c/p\u003e\n\u003cp dir=\"LTR\"\u003e\u003cstrong\u003eDisclaimer/\u003c/strong\u003e\u003cstrong\u003ePublisher\u0026apos;s Note:\u003c/strong\u003e The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eVillaverde, P. et al. Hypothetical impact of the Mexican front-of-pack labeling on intake of critical nutrients and energy. \u003cem\u003eJ. Health Popul. Nutr.\u003c/em\u003e \u003cb\u003e42\u003c/b\u003e (1), 124. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s41043-023-00462-7\u003c/span\u003e\u003cspan address=\"10.1186/s41043-023-00462-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023). PMID: 37941048; PMCID: PMC10631037.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBassetti, E., Khosravi, A. \u0026amp; Pries, A. M. Prevalence of Front-of-Pack Warning Signs among Commercial Complementary Foods in Seven High and Upper Middle-Income Countries. \u003cem\u003eNutrients\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e (7), 1629. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu15071629\u003c/span\u003e\u003cspan address=\"10.3390/nu15071629\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023). PMID: 37049470; PMCID: PMC10096843.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoark, E. K., Rehm, C. D. \u0026amp; Sherry, C. L. Potential for Front of Pack Labeling Exposure to Impact US Dietary Choices: A Population-Based Cross-Sectional Study Using NHANES 2017\u0026ndash;2018. \u003cem\u003eNutrients\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e (14), 2995. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu14142995\u003c/span\u003e\u003cspan address=\"10.3390/nu14142995\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022). PMID: 35889951; PMCID: PMC9316916.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSarda, B. et al. Complementarity between the updated version of the front-of-pack nutrition label Nutri-Score and the food-processing NOVA classification. \u003cem\u003ePublic. Health Nutr.\u003c/em\u003e \u003cb\u003e27\u003c/b\u003e (1), e63 (2024). PMID: 38297466; PMCID: PMC10897572.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMulligan, C., Lee, J. J., Vergeer, L., Ahmed, M. \u0026amp; L'Abb\u0026eacute;, M. R. Evaluating the potential implications of canadian front-of-pack labelling regulations in generic and branded food composition databases. \u003cem\u003eBMC Public. Health\u003c/em\u003e. \u003cb\u003e22\u003c/b\u003e (1), 1866. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12889-022-14269-4\u003c/span\u003e\u003cspan address=\"10.1186/s12889-022-14269-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022). PMID: 36203128; PMCID: PMC9535871.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShin, S. et al. A Randomized Controlled Study to Test Front-of-Pack (FOP) Nutrition Labels in the Kingdom of Saudi Arabia. \u003cem\u003eNutrients\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e (13), 2904. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu15132904\u003c/span\u003e\u003cspan address=\"10.3390/nu15132904\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023). PMID: 37447230; PMCID: PMC10343722.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, J. J. et al. Examining the diet quality of Canadian adults and the alignment of Canadian front-of-pack labelling regulations with other front-of-pack labelling systems and dietary guidelines. Front Public Health. ;11:1168745. doi: 10.3389/fpubh.2023.1168745. Erratum in: Front Public Health. 2024;12:1448853. (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fpubh.2024.1448853\u003c/span\u003e\u003cspan address=\"10.3389/fpubh.2024.1448853\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 37427256; PMCID: PMC10326271.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShin, S., Puri, J. \u0026amp; Finkelstein, E. A randomized trial to evaluate the impact of Singapore's forthcoming Nutri-grade front-of-pack beverage label on food and beverage purchases. \u003cem\u003eInt. J. Behav. Nutr. Phys. Act.\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e (1), 18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12966-023-01422-4\u003c/span\u003e\u003cspan address=\"10.1186/s12966-023-01422-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023). PMID: 36793058; PMCID: PMC9930047.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSarda, B. et al. Consistency of the Initial and Updated Version of the Nutri-Score with Food-Based Dietary Guidelines: A French Perspective. \u003cem\u003eJ. Nutr.\u003c/em\u003e \u003cb\u003e154\u003c/b\u003e (3), 1027\u0026ndash;1038 (2024). Epub 2024 Feb 2. PMID: 38311063; PMCID: PMC10942857.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller, C., Kay, E., Dono, J. \u0026amp; Ettridge, K. Impacts of sugar and sweetener warning labels on substitution between sugar- and non-sugar-sweetened beverages in a non-hypothetical selection task. \u003cem\u003eBMC Med.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e (1), 541. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12916-024-03740-1\u003c/span\u003e\u003cspan address=\"10.1186/s12916-024-03740-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). PMID: 39558314; PMCID: PMC11574984.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller, C., Ettridge, K., Kay, E. \u0026amp; Dono, J. What about 100% juice and non-sugar sweeteners? A national study of support for taxes, labelling and marketing bans applied to sugary drinks, non-sugar sweetened beverages and 100% juice in Australia. \u003cem\u003eAust N Z. J. Public. Health\u003c/em\u003e. \u003cb\u003e49\u003c/b\u003e (3), 100238 (2025). Epub 2025 Apr 29. PMID: 40307103.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller, C. et al. Warning labels for sugar-sweetened beverages and fruit juice: evaluation of 27 different labels on health effects, sugar content, energy and exercise equivalency. \u003cem\u003ePublic. Health\u003c/em\u003e. \u003cb\u003e230\u003c/b\u003e, 138\u0026ndash;148 (2024). Epub 2024 Mar 27. PMID: 38547760.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBassetti, E. et al. Benchmarking the nutrient composition and labelling practices of commercially produced ready-to-eat pur\u0026eacute;es and meals for older infants and young children across seven Southeast Asian countries. \u003cem\u003eMatern Child. Nutr.\u003c/em\u003e \u003cb\u003e19\u003c/b\u003e (Suppl 2(Suppl 2), e13585. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/mcn.13585\u003c/span\u003e\u003cspan address=\"10.1111/mcn.13585\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023). PMID: 38092377; PMCID: PMC10719058.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSagaceta-Mejia, J. et al. Purchase behavior reported by Mexican adults after implementing the front-of-pack warning labels. \u003cem\u003eSalud Publica Mex\u003c/em\u003e. \u003cb\u003e67\u003c/b\u003e (5 (sept-oct)), 447\u0026ndash;454 (2025). Spanish. doi: 10.21149/16253. PMID: 41150992.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoukid, F. \u0026amp; Castellari, M. Food and Beverages Containing Algae and Derived Ingredients Launched in the Market from 2015 to 2019: A Front-of-Pack Labeling Perspective with a Special Focus on Spain. \u003cem\u003eFoods\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e (1), 173. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/foods10010173\u003c/span\u003e\u003cspan address=\"10.3390/foods10010173\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021). PMID: 33467009; PMCID: PMC7830845.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKroker-Lobos, M. F. et al. Efficacy of front-of-pack warning label system versus guideline for daily amount on healthfulness perception, purchase intention and objective understanding of nutrient content of food products in Guatemala: a cross-over cluster randomized controlled experiment. \u003cem\u003eArch. Public. Health\u003c/em\u003e. \u003cb\u003e81\u003c/b\u003e (1), 108. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13690-023-01124-0\u003c/span\u003e\u003cspan address=\"10.1186/s13690-023-01124-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023). PMID: 37328782; PMCID: PMC10273755.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQuevedo, K. L. et al. Sociodemographic differences in nutrition labels effect on Chilean and Mexican youth. \u003cem\u003eRev. Panam. Salud Publica\u003c/em\u003e. \u003cb\u003e48\u003c/b\u003e, e33 (2024). PMID: 38576841; PMCID: PMC10993807.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHashem, K. Towards a more nuanced understanding of policies that lead to food reformulation for a food system change. Proc Nutr Soc. 2025 Dec 26:1\u0026ndash;23. doi: 10.1017/S002966512510205X. Epub ahead of print. PMID: 41449698.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTomaz, L. A. et al. From the most to the least flexible nutritional profile: Classification of foods marketed in Brazil according to the Brazilian and Mexican models. \u003cem\u003eFront. Nutr.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, 919582 (2022). PMID: 36204372; PMCID: PMC9531871.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller, C. et al. Qualitative insights into Australian consumers' views for and against government action on sugary drinks. Public Health Res Pract. ;31(2):30122003. (2021). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.17061/phrp30122003\u003c/span\u003e\u003cspan address=\"10.17061/phrp30122003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 34104932.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCrosbie, E. et al. A policy study on front-of-pack nutrition labeling in the Americas: emerging developments and outcomes. \u003cem\u003eLancet Reg. Health Am.\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e, 100400 (2022). PMID: 36844016; PMCID: PMC9950544.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartini, D. et al. Can Front-of-Pack Labeling Encourage Food Reformulation? A Cross-Sectional Study on Packaged Bread. \u003cem\u003eFoods\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e (22), 3535. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/foods13223535\u003c/span\u003e\u003cspan address=\"10.3390/foods13223535\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). PMID: 39593951; PMCID: PMC11593041.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMialon, M. Front-of-pack nutrition labeling: ultra-processed foods are the elephant in the room. Lancet Reg Health Am. ;18:100442. (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.lana.2023.100442\u003c/span\u003e\u003cspan address=\"10.1016/j.lana.2023.100442\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 36844010; PMCID: PMC9950663.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDevaux, M., Aldea, A., Lerouge, A., Vuik, S. \u0026amp; Cecchini, M. Establishing an EU-wide front-of-pack nutrition label: Review of options and model-based evaluation. \u003cem\u003eObes. Rev.\u003c/em\u003e \u003cb\u003e25\u003c/b\u003e (6), e13719. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/obr.13719\u003c/span\u003e\u003cspan address=\"10.1111/obr.13719\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). Epub 2024 Feb 7. PMID: 38326224.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJulia, C., Gokani, N., Hercberg, S. \u0026amp; Garde, A. Ten years of Nutri-Score front-of-pack nutrition labelling in Europe. \u003cem\u003eNat. Food\u003c/em\u003e. \u003cb\u003e6\u003c/b\u003e (3), 239\u0026ndash;243. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s43016-025-01141-y\u003c/span\u003e\u003cspan address=\"10.1038/s43016-025-01141-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025). Epub 2025 Mar 13. PMID: 40082656.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMerz, B. et al. Nutri-Score 2023 update. \u003cem\u003eNat. Food\u003c/em\u003e. \u003cb\u003e5\u003c/b\u003e (2), 102\u0026ndash;110. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s43016-024-00920-3\u003c/span\u003e\u003cspan address=\"10.1038/s43016-024-00920-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). Epub 2024 Feb 14. PMID: 38356074.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKelly, B., Ng, S. H., Carrad, A. \u0026amp; Pettigrew, S. The Potential Effectiveness of Front-of-Pack Nutrition Labeling for Improving Population Diets. \u003cem\u003eAnnu. Rev. Nutr.\u003c/em\u003e \u003cb\u003e44\u003c/b\u003e (1), 405\u0026ndash;440. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1146/annurev-nutr-011224-030917\u003c/span\u003e\u003cspan address=\"10.1146/annurev-nutr-011224-030917\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). Epub 2024 Aug 12. PMID: 38857543.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYin, X. et al. Stakeholder Network Analysis for Front-of-Pack Labeling in China. \u003cem\u003eFront. Nutr.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, 871062 (2022). PMID: 35677544; PMCID: PMC9169156.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong, J. et al. Impact of color-coded and warning nutrition labelling schemes: A systematic review and network meta-analysis. \u003cem\u003ePLoS Med.\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e (10), e1003765. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pmed.1003765\u003c/span\u003e\u003cspan address=\"10.1371/journal.pmed.1003765\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021). PMID: 34610024; PMCID: PMC8491916.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFialon, M., Nabec, L. \u0026amp; Julia, C. Legitimacy of Front-of-Pack Nutrition Labels: Controversy Over the Deployment of the Nutri-Score in Italy. \u003cem\u003eInt. J. Health Policy Manag\u003c/em\u003e. \u003cb\u003e11\u003c/b\u003e (11), 2574\u0026ndash;2587. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.34172/ijhpm.2022.6127\u003c/span\u003e\u003cspan address=\"10.34172/ijhpm.2022.6127\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022). Epub 2022 Feb 20. PMID: 35219283; PMCID: PMC9818094.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu, F. et al. Prepackaged foods healthiness ranking index: developing front-of-pack labeling to facilitate the assessment and validation of healthiness rankings of prepackaged foods in China. \u003cem\u003eBMC Med.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e (1), 402. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12916-024-03600-y\u003c/span\u003e\u003cspan address=\"10.1186/s12916-024-03600-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). PMID: 39300429; PMCID: PMC11414037.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePettigrew, S. et al. A review of front-of-pack nutrition labelling in Southeast Asia: Industry interference, lessons learned, and future directions. \u003cem\u003eLancet Reg. Health Southeast. Asia\u003c/em\u003e. \u003cb\u003e3\u003c/b\u003e, 100017 (2022). PMID: 37384259; PMCID: PMC10305914.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChung, A., Hatzikiriakidis, K., Martino, F. \u0026amp; Skouteris, H. Characterising Parent-Appeal Marketing on Foods for Children: A Scoping Review. \u003cem\u003eCurr. Nutr. Rep.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e (3), 393\u0026ndash;398. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s13668-024-00559-3\u003c/span\u003e\u003cspan address=\"10.1007/s13668-024-00559-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024). Epub 2024 Jun 27. PMID: 38935250; PMCID: PMC11327212.\u003c/span\u003e\u003c/li\u003e \u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Front-of-pack labeling, Nutrient profiling, Reformulation incentives, Artificial sweeteners, Clean label, Signal credibility, Discrete choice experiment","lastPublishedDoi":"10.21203/rs.3.rs-8573773/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8573773/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFront-of-pack (FOP) nutrient grading systems are intended to shift demand toward healthier products and to incentivize reformulation. In beverages, however, favorable grades are often achieved through sugar reduction via artificial sweetener substitution, potentially creating a credibility conflict between an algorithmic \"health\" signal and ingredient cues that consumers interpret as artificial or non\u0026ndash;clean label. This study tests whether the demand-side effectiveness of nutrient grades is conditional on reformulation pathways. We conducted a discrete choice experiment with 2,736 Chinese adults (12 choice tasks; opt-out included) in which generic 330 mL carbonated soft drinks varied by Nutri-Grade, sweetener strategy, flavor, and price. Random parameters logit models show a robust grade\u0026ndash;ingredient interaction consistent with conditional signaling: the Grade A premium is substantially attenuated when paired with the artificial sweetener blend, flattening predicted choice gains from grade improvements. In monetary terms, willingness-to-pay for upgrading from Grade C to Grade A is +\u0026thinsp;2.31 RMB under cane sugar but becomes statistically indistinguishable from zero (+\u0026thinsp;0.27 RMB) under artificial sweeteners, implying a valuation erosion of about 2.04 RMB. Latent class models reveal strong heterogeneity: a large \"clean-label purist\" segment (44.2%) drives grade discounting under artificial sweeteners, whereas \"algorithmic health seekers\" (32.5%) respond strongly to grades with minimal discounting. The findings indicate that nutrient grades function as conditional policy signals whose effectiveness depends on the reformulation strategies used to obtain favorable scores, with implications for how FOP grading policies are designed and evaluated in categories where additive-salient reformulation is prevalent.\u003c/p\u003e","manuscriptTitle":"Reformulation pathways shape the impact of front-of-pack nutrient grades: A discrete choice experiment on sweetener substitution","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-21 07:23:55","doi":"10.21203/rs.3.rs-8573773/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"aad1ebfb-7106-40f7-9d43-269b7b5304cc","owner":[],"postedDate":"January 21st, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":61435701,"name":"Physical sciences/Mathematics and computing"},{"id":61435702,"name":"Biological sciences/Psychology"},{"id":61435703,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2026-02-26T07:56:55+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-21 07:23:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8573773","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8573773","identity":"rs-8573773","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}