CF labeling is an instantly recognizable moral cue, signaling that no animal suffered during the product’s development and appealing to consumers’ other-regarding concerns (Winders, 2006). Empirical research consistently demonstrates that this signal resonates with individuals who value animal welfare intrinsically. In cosmetics, for instance, respondents who notice a CF logo not only form more favorable attitudes but also report stronger purchase intentions, an effect that Wuisan and Februadi (2022) attribute to the importance that such consumers attach to safeguarding animals. Similar evidence is presented by Grappe et al. (2021), who demonstrate that CF logos harmonize ECs with self-presentation motives, yielding a distinctly positive affective response. (Munteanu and Pagalea, 2014; Lewis et al., 2016).

These findings accord with moral-heuristic theory: when shoppers face time or information constraints, they rely on easily processed “ethical shortcuts.” (De Pelsmacker et al., 2005). Sheehan and Lee (2014) show that a CF icon serves precisely this purpose, enabling quick alignment between the act of buying and the buyer’s moral identity. Such alignment is fundamentally altruistic because the primary beneficiary is a non-human other.

More broadly, sustainable consumption studies confirm that the desire to act in the interest of others—whether humans, animals, or the ecosystem—constitutes a pivotal motive behind ethical purchases (Waris et al., 2021; Schuitema and de Groot, 2015). Accordingly, a CF label should do more than enhance product appeal; it should kindle AM by reminding consumers that their choice can prevent harm to sentient beings.

Influencers constitute a new class of value transmitters whose authenticity, narrative intimacy, and algorithm-enabled reach make them unusually effective at mobilizing other-regarding motives. Parasocial-relationship theory predicts that when a trusted content creator publicizes an ethical stance, followers are primed to adopt that stance as part of their own moral identity; recent work confirms the prediction across multiple contexts. For green cosmetics, social-media advocacy elevates subjective norms and, through them, AM to support CF brands (Pop et al., 2020). In a broader sustainability context, altruistic—as well as egoistic—motives mediate the effect of influencer messages on purchase intentions (Kumar and Pandey, 2023). Studies in high-credibility contexts show that trust in the messenger amplifies this moral contagion (Chavda and Chauhan, 2024).

The persuasive arc does not end with mere attitude change: altruistically motivated followers go on to circulate ethical content themselves, extending the influencer’s reach through peer sharing (Plume and Slade, 2018). Such network effects help explain why influencer campaigns are now a cornerstone of socially responsible branding strategies (Lokithasan et al., 2019). They also clarify why altruistic concern predicts online purchase intention for ethically positioned products, especially among digital-native cohorts (Traymbak et al., 2022). This evidence supports the proposition that influencer advocacy catalyzes AM, turning moral sentiment into market behavior.

CSR initiatives serve as moral signals that help consumers assess a firm’s alignment with their ethical standards (Sen and Bhattacharya, 2001; Chen and Kim, 2019). When the signal is interpreted as authentic and other-focused, the firm is reclassified from a market actor to a “moral in-group” partner, prompting a self-transcendent desire to advance shared prosocial aims—an archetypal form of AM (Bhattacharya and Sen, 2004). Crucially, this motivational lift hinges on perceived sincerity rather than the mere presence of a CSR program: high-fit, proactively communicated initiatives strengthen moral identification, whereas low-fit or transparently profit-driven efforts erode it (Becker-Olsen et al., 2005). Empirical work corroborates the mechanism across sectors: Consumers who ascribe altruistic motives to CSR in services and retailing report warmer brand evaluations and a stronger intention to reward the firm (Pérez and Bosque, 2015). Authentic, other-oriented messaging elevates moral emotions and measurably heightens AM (Wong and Dhanesh, 2017; Ji and Jan, 2019); influencer-mediated CSR endorsements likewise increase followers’ readiness to emulate prosocial behavior (Cheng et al., 2021). Meta-analytic evidence further shows that congruence between a consumer’s altruistic values and a focal CSR cause amplifies attitudinal and behavioral outcomes (Ribeiro et al., 2022). Together, these findings suggest that a credible CSR image operates as a psychological catalyst, converting ethical appreciation into an intrinsic willingness to act for the welfare of others.

PF sets the economic “permission structure” that determines whether moral motives can be enacted in the marketplace. Building on dual-entitlement theory, consumers simultaneously evaluate (a) their own right to a reference price and (b) the firm’s right to a reasonable profit (Kahneman et al., 1986). When a CF option is priced within this reference zone—or only slightly above it as a justifiable “moral premium”—the economic cost of acting ethically is perceived as tolerable, preserving the intrinsic warm-glow reward that follows prosocial choices (Curth et al., 2024). Equity-based appraisals then remain neutral or positive, allowing symbolic and social cues (such as a logo, influencer advocacy, or CSR) to reach the motivation stage unimpeded. If, however, the price breaches the entitlement boundary, the transaction is re-framed from “shared virtue” to “moral surcharge,” shifting attention from animal welfare to self-protection and heightening loss-aversion concerns (Bolton et al., 2003).

Crucially, fairness evaluations operate upstream, filtering how strongly AM translates into downstream cognitive and behavioral processes. Neuro-imaging work shows that prices deemed unjust extinguish reward-system activation triggered by prosocial cues long before a purchase decision is finalized (Chang et al., 2011). Survey and diary studies also reveal that consumers confronted with perceived overpricing quickly downgrade their moral intent to mere approval, preserving their self-concept while withholding action (Xia and Monroe, 2010; Chong et al., 2021; Maxwell, 2005). In short, PF functions as a boundary stimulus, not because it directly sparks or suppresses compassion, but because it modulates the pathway through which existing altruistic motives become concrete, CF purchases. This boundary role provides the conceptual scaffold for the moderated-mediation hypotheses (H13–H15) developed later in the paper without pre-empting their specific outcome-focused logic.

Altruistic motivation (AM)—defined as an internalized concern for the welfare of others—activates personal moral norms and a felt obligation to reduce harm (Stern et al., 1995). When such motivation is salient, CF labels become powerful moral affordances, enabling consumers to align marketplace choices with their prosocial values (Schuitema and de Groot, 2015; Magano et al., 2022). Experimental and survey evidence converge on this mechanism: psychosocial values that prioritize animal well-being enhance attitudes toward CF cosmetics and, in turn, purchase intentions (Grappe et al., 2021; Wuisan and Februadi, 2022). Longitudinal and cross-sectional studies further show that these attitudinal gains translate into behavior; altruistic concern predicts both stated intent and verified CF purchases across food, apparel, and beauty contexts (Jaiswal and Kant, 2018; Yazdanpanah and Forouzani, 2015; Prakash et al., 2024). Within the Theory of Planned Behavior, altruism serves as a core antecedent of pro-environmental attitudes (Huang et al., 2018), while value–belief–norm research suggests that moral obligation mediates the transition from intent to action (Chaudhary, 2018). Marketplace studies echo the finding that brands that embed altruistic cues in their positioning elicit stronger engagement and loyalty (Boccadoro et al., 2021). Additionally, social media communities amplify altruistic motives through peer endorsement and normative reinforcement (Kumar and Pandey, 2023). The evidence portrays altruistically motivated consumers as viewing CF products not merely as functional goods but as vehicles for enacting and reinforcing their ethical identity (Aggarwal et al., 2024), thereby increasing the likelihood—and persistence—of CF buying.

Altruistic orientations have a positive and significant impact on consumers’ attitudes toward ethical consumption (Oh and Yoon, 2014). This effect arises because individuals committed to altruistic values tend to internalize ethical consumption as part of their self-concept (Culiberg and Bajde, 2013; Chi, 2022; Bae and Yan, 2018). Even when deliberative processes obscure altruistic motives, intuitive, other-regarding reactions often yield ethically favorable decisions (Zhong, 2011). Accordingly, when consumers act on altruistic goals—such as reducing harm to animals or protecting vulnerable beings—their EC is likely to intensify (Le-Hoang, 2025; Chang and Chuang, 2020). They not only perceive cruelty-free options as preferable but also experience a stronger affective response to any potential moral transgressions (Schwartz, 1977). These insights lead us to the following hypothesis:

Ethical concern (EC)—the cognitive-affective appraisal that one’s consumption choices carry moral consequences for sentient beings—acts as a pivotal catalyst in CFBB. Laboratory and field studies demonstrate that when empathy is considered in the decision-making process, consumers consistently prioritize products that prevent animal harm (Grappe et al., 2021; Michaelidou and Hassan, 2007; Fox and Ward, 2008). This effect is particularly pronounced among individuals with a pronounced internal locus of control, who are more likely to act when they believe their purchase can influence ethical outcomes (Lim et al., 2019; Dasunika and Gunathilake, 2021; Dissanayake, 2022). Experimental evidence further indicates that prompting moral reflection increases preference for ethically marketed goods (Peloza et al., 2013), while information on animal cruelty redirects demand toward CF options (Gunther et al., 2023). Collectively, these findings position EC as a behavioral driver rather than a mere attitude.

Choosing CF products enables consumers to establish a value alignment that extends beyond utilitarian exchange and serves as a symbolic self-presentation (White and Argo, 2011). Identity-based consumption theory posits an inspiration sequence: a brief “being inspired by” affective spark on recognizing a value match, followed by a “being inspired to” motivation to display that value publicly (Thrash and Elliot, 2003; Reed et al., 2012). In CF purchases, the act itself signals “I care about animal welfare,” supplying a potent “inspiration-by” trigger. Empirical evidence confirms that value-laden consumption catalyzes visual and behavioral self-expressions: admirers of ethical brands convert their purchases into identity markers that reinforce personal narratives and social images (Schau and Gilly, 2003; Schuitema and de Groot, 2015; van der Westhuizen and Kuhn, 2024). Likewise, Magano et al. (2022) demonstrate that altruism- and responsibility-based attitudes toward CF cosmetics encourage consumers to utilize their purchasing decisions as a platform for ethical advocacy. Together, these findings support the expectation that CFBB elicits SEI.

Affiliation inspiration (AI) describes the brief but intense sense of “we-ness” that arises when an individual’s action resonates with a shared ethical norm or community identity (Schau et al., 2009). According to social identity theory, internalizing group values strengthens one’s sense of belonging (Bhattacharya and Sen, 2003). By choosing a CF product, consumers visibly signal membership in an “animal-friendly” community, igniting their AI. Moreover, generative cues—such as imagery emphasizing benefits for future generations—reinforce this communal bond by conveying that “our actions today serve the well-being of tomorrow’s members” (Ma and Xing, 2024). The resulting warm-glow effect further amplifies collective responsibility and the desire to unite with like-minded others (Tezer and Bodur, 2020). Taken together, these processes support the following hypothesis:

Psychological empowerment theory frames behavioral empowerment as the momentary surge of competence and agency that follows an act perceived as personally efficacious (Zimmerman, 1995). At its core are three mutually reinforcing routes—cognitive, emotional, and behavioral—through which an action nurtures intrinsic motivation (Thomas and Velthouse, 1990). CF purchasing activates each route. Cognitively, the shopper apprehends concrete animal welfare gains, imbuing the choice with meaning and personal relevance. Emotionally, the “warm-glow” effect accompanying prosocial action fortifies moral self-confidence (Chen et al., 2020). Behaviorally, enacting an ethical preference confirms one’s ability to effect change, thereby elevating self-efficacy and encouraging sustainable decisions in the future (Bandura, 1997). Empirical work shows that such empowerment experiences arise whenever technology or information grants users new capacities, whether through e-government portals (Li and Gregor, 2011) or peer-generated reviews that expand choice autonomy (Hu and Krishen, 2019). Accordingly, CF consumption should reliably kindle BEI by informing, energizing, and mobilizing consumers in a single, self-reinforcing episode.

A cross-sectional sample of N = 624 adult consumers residing in Türkiye was recruited via a professional online panel provider. We employed a stratified quota sampling strategy to ensure proportional representation across key demographic strata: gender (52% female, 48% male), age (18–24 = 18%, 25–34 = 37%, 35–44 = 29%, 45 + = 16%), and NUTS-1 region (Marmara, Central Anatolia, Aegean, Mediterranean, Black Sea, Eastern, and Southeastern Anatolia). This approach was chosen because purchase motivations, price sensitivity, and social influence cues are known to vary systematically by demographic segment. Stratification reduces sampling error, while quota controls mitigate overrepresentation, thereby enhancing the generalizability of findings to Türkiye’s adult population (Dillman et al., 2014).

Participants were eligible if they (a) purchased personal-care or household products at least once per month and (b) used a smartphone for shopping-related activities—criteria selected to reflect real-life exposure to CF product cues. An a priori power analysis using G*Power 3.1 (f² = 0.02, α = 0.05, power = 0.95, 10 predictors) indicated a minimum required sample size of 474; the final sample comfortably exceeded this threshold. Of 721 invitations distributed, 654 responses were received (90.7% response rate), and 30 were excluded due to inattentive responding or survey completion times below one-third of the median, resulting in 624 valid responses. As shown in Table 2, sample demographics closely mirror national census figures, further supporting the external validity of the dataset.

All latent constructs were operationalized with multi-item, reflective scales adapted from prior peer-reviewed research. Scale wording was first adjusted to the CF products context, then subjected to a double back-translation procedure (English ↔ Turkish) to secure semantic equivalence. A panel of three marketing scholars and two industry practitioners evaluated the content validity, resulting in minor lexical refinement.

Items for SEI and AI (3 each) were adapted from Schau et al. (2009) and Venn et al. (2017). AM (4 items) was drawn from Goldsmith et al. (2000) and Prakash et al. (2024). The eco-label (CF logo) perception scale (3 items) followed by Nittala (2014) and Song et al. (2020). EC (4 items) relied on Suphasomboon and Vassanadumrongdee (2022). The PCSR image (7 items) was adapted from Achabou and Ho’s environmental CSR scale and tailored to the CF domain (Achabou, 2020; Ho, 2017; Huang et al., 2022). PF (6 items) combined wording from Petrick (2002), Martin et al. (2009), and Chung and Petrick (2013). The five-item SMI credibility scale was customised by De Veirman et al. (2017). CFBB (6 items) was adapted from Khare (2015), and behavioral-empowerment inspiration (3 items) from Speer and Peterson (2000), Christens (2012), and Li et al. (2021). The original scale items are listed in Appendix Table A.

All items were anchored on a seven-point Likert scale (1 = “strongly disagree” to 7 = “strongly agree”). A pilot test with 80 respondents confirmed readability and yielded satisfactory internal consistency reliabilities (Cronbach’s α ≥ 0.78). Pilot data also showed no cross-loading above 0.30 in an exploratory factor analysis, supporting preliminary discriminant validity. Final measurement properties (indicator loadings, AVE, CR, α) are reported in Table 3 and meet recommended thresholds for PLS-SEM. Common-method bias was assessed ex-ante through proximal item placement and ex-post via the full collinearity VIF test; all latent VIFs were < 3.3, indicating no substantial bias.

As shown in Table 3, all constructs satisfied established reliability and convergent validity benchmarks (Hair et al., 2024). Outer loadings ranged from 0.717 to 0.980, comfortably above the 0.70 guideline, except for two indicators (CFL2 = 0.717; SMI5 = 0.702) that were retained for content coverage. Cronbach’s α coefficients (0.723–0.969) and composite reliabilities (0.828–0.980) exceeded the 0.70 threshold, indicating strong internal consistency (Nunnally and Bernstein, 1994). Average variance extracted (AVE) values (0.570–0.942) surpassed the 0.50 criterion, confirming convergent validity. Variance-inflation factors for all indicators (1.26–3.44) fell well below the conservative threshold of 5, suggesting no concerns about multicollinearity (Kock and Lynn, 2012). Collectively, these statistics demonstrate that the measurement model is both reliable and convergent, providing a sound basis for subsequent structural analysis.

As displayed in Table 4, the square roots of AVE values (0.755–0.971; diagonal) exceed every inter-construct correlation, including the largest observed off-diagonal value (0.825 between EC and SMIs). This pattern satisfies the Fornell–Larcker requirement that a latent construct share more variance with its indicators than with any other construct (Fornell and Larcker, 1981; Hair et al., 2024).

Table 5 reports Heterotrait–Monotrait ratios ranging from 0.227 to 0.793, well below the conservative 0.85 threshold and the liberal 0.90 guideline (Henseler et al., 2015; Hair et al., 2024). Hence, every construct remains empirically distinct; even the highest value—between CFL and CFBB—does not threaten discriminant validity.

Variance-inflation factors for endogenous predictors (Table 6) range from 1.26 to 2.97. The highest values—2.97 for PCSR Image, 2.96 for SMIs, and 2.78 for EC—remain comfortably beneath both the classical ceiling of 5 (Hair et al., 2024) and the stricter lateral-collinearity safeguard of 3.3 (Kock and Lynn, 2012). The interaction term (PF × AM) shows an ideal VIF of 1.58, indicating near orthogonality.

The f² matrix in Table 7 highlights the model’s parsimony. Large effects (e.g., 0.862, 0.872, 0.706, 0.697) indicate that these predictors explain a substantial proportion of variance in their respective outcomes. Medium contributions around 0.10–0.14 add incremental explanatory power, whereas the near-zero coefficients (0.020, 0.011, 0.009) denote links rendered redundant by full mediation—once the theorized mediator is introduced, the direct path adds no further R² (Cohen, 1988). This pattern—strong drivers where theory predicts them and trivial direct effects where mediation is posited—confirms the structural model’s coherence and empirical efficiency.

The structural model exhibits strong explanatory and predictive power (Table 8). CFBB (R² = 0.736) and SEI (0.711) approach the “substantial” threshold, while AI (0.556) and AM (0.496) fall comfortably within the “moderate” range. Even the lowest R² values—BEI (0.418) and EC (0.417)—exceed the 0.25 benchmark, indicating that the model accounts for a meaningful proportion of variance across all endogenous constructs (Hair et al., 2024; Henseler et al., 2015). Complementing this explanatory strength, predictive-relevance diagnostics further support model robustness: Stone–Geisser Q² values range from 0.421 to 0.808, well above zero, confirming out-of-sample accuracy. Likewise, RMSA coefficients fall between 0.440 and 0.576, reflecting moderate and proportionate residual error relative to construct complexity (Shmueli et al., 2016), thereby affirming the model’s empirical adequacy for both theoretical insight and practical inference.

Global fit indices reinforce overall adequacy. As shown in Table 8, the model’s SRMR of 0.084 is below the conservative threshold of 0.10, and the NFI of 0.878 approaches the recommended benchmark of 0.90 (Hair et al., 2024), indicating that the reproduced covariance matrix closely matches the observed data.

Bootstrapped path estimation (5,000 resamples) produced the coefficient matrix summarized in Table 9 and visualized in Figure 2. Overall, the model accounts for 42–74% of variance across its six endogenous constructs, providing a robust platform for hypothesis testing.

The empirical results align closely with the study’s theorization. First, the three exogenous cues—CF labeling, SMI advocacy, and PCSR image—each exerts a positive, significant, and theoretically coherent impact on AM (H1–H3). Among them, CSR is the most potent predictor (β = 0.539, p < 0.001), which is consistent with the stakeholder-identification view that consumers internalize socially responsible signals as moral self-relevance (Bhattacharya and Sen, 2004). The smaller but meaningful coefficients for labeling (β = 0.282) and SMIs endorsement (β = 0.160) suggest that tangible on-pack cues and parasocial persuasion work in tandem. However, corporate deeds weigh more heavily than words or badges.

AM, in turn, operates exactly as posited: it promotes EC (β = 0.646). It directly encourages CF buying (β = 0.339), supporting value-belief-norm logic and the moral extension of the Theory of Planned Behavior (Ajzen, 1991). EC partially carries this effect forward (H6; β = 0.243), yielding a layered motivational pathway that culminates in purchase. The strength and significance of these links validate H4–H6 and justify the sequential-mediation tests. Downstream, CF buying acts as a springboard for three post-purchase inspirations—self-expression (H7, β = 0.843), affiliation identity (H8, β = 0.745) and behavioral empowerment (H9, β = 0.647)—all far exceeding the “large-effect” benchmark in PLS-SEM.

The sequential mediations (H10–H12) are all significant (p ≤ 0.05). Each indirect chain, from the three upstream cues through AM and EC to buying behavior, registers, demonstrating that the cues work primarily by elevating moral motives, rather than bypassing them. This finding reinforces the argument that CF consumption is less an impulse purchase than a moralized decision requiring internalized justification.

The moderated-mediation hypotheses (H13–H15) are supported: the interaction of PF with AM strengthens the indirect effect of buying behavior on all three inspiration outcomes. Fair pricing, therefore, helps motivated consumers translate moral intent into action and, in turn, derive richer psychological rewards.

Every hypothesis is statistically upheld, and effect sizes follow theoretical expectations—strongest for value-based links, moderate for informational cues, and weakest (yet still significant) for direct shortcuts explicitly treated as mediated. The pattern reinforces the coherence and practical relevance of the proposed causal chain from corporate and social signals to individual motives, behavior and identity-building outcomes.

The qualitative phase grounded in in-depth interviews complemented the quantitative PLS-SEM findings developed within the S-O-R framework. Recruitment, scheduling, and completion of the interviews took 3 months (March–May 2025). We initially aimed for 20–25 in-depth interviews to balance analytic depth with diversity in CF involvement, age, gender, and occupational background. Participants were recruited purposively to capture variation in exposure to CF products and ethical consumption discourse (e.g., long-term CF users, occasional CF buyers, and consumers who were only recently aware of CF labels). Saturation was tracked iteratively across this heterogeneous sample: after 20 interviews, no substantively new codes emerged, and two additional interviews were conducted with participants from underrepresented profiles, confirming the stability of the thematic structure, resulting in a final sample of 22 interviews. Using purposeful sampling, 22 participants—drawn from diverse socio-demographic backgrounds and exhibiting varying levels of experience with CF consumption—took part in semi-structured interviews lasting 45–75 min each (Patton, 2015). The semi-structured interview guide comprised 14 questions addressing the study’s focal stimuli (CFL presence, SMI advocacy, PCSR image, PF), organism-level mediators (AM, EC), and response-level outcomes (CFBB, SEI, AI, BEI); the full guide is available in Appendix Table B.

Table 10 summarises the demographic and experiential profiles of the 22 interviewees. The ages ranged from the early 20s to the early 40s, and the sample achieved an intentionally balanced gender representation—fourteen women and eight men—reflecting the slight female tilt typically reported in CFproduct markets. Educational attainment ranged from high school diplomas to doctoral degrees. At the same time, occupations included knowledge-intensive roles (e.g., UX designer, data analyst, university lecturer, beauty specialist) and service-sector positions (e.g., barista, retail manager). This breadth ensured the inclusion of consumers with disparate disposable incomes and workplace cultures, factors that influence ethical purchase priorities.

The qualitative dataset comprised verbatim transcripts totaling approximately 132,000–135,000 words, all of which were imported into MAXQDA 2022 for systematic analysis. We followed Braun and Clarke’s (2006) six-phase thematic analysis protocol, beginning with familiarization and inductive line-by-line coding. To preserve conceptual alignment with our quantitative model, coding remained sensitized to pre-specified constructs (deductive) while allowing novel insights to emerge organically (inductive) (Nowell et al., 2017).

Saturation was tracked iteratively; after 20 interviews, no novel first-order codes appeared across two consecutive transcripts, yet two additional interviews were conducted to confirm redundancy (Malterud et al., 2016). The initial pool of 148 open codes was refined through code reconciliation and merged into 63 focused codes. These were then organized into 21 sub-themes and synthesized into seven integrative analytical themes (see Table 11). Together, these themes furnish a rich narrative that deepens understanding of how CF purchasing decisions emerge from the interplay of situational cues, moral motivations, and identity-driven outcomes.

To ensure methodological rigor in line with COREQ guidelines (Tong et al., 2007), we maintained a detailed audit trail documenting every stage of theme development. Reflective memos capture analytical decisions and researcher reflexivity (Nowell et al., 2017). The qualitative dataset comprised verbatim transcripts, which were analysed using reflexive thematic analysis. An initial codebook was collaboratively developed by two researchers, combining theory-informed sensitising concepts with inductive codes derived from a close reading of a subset of transcripts. Researchers independently coded this subset, compared coding decisions, and resolved discrepancies through discussion, resulting in a refined, shared codebook. Inter-coder reliability was assessed by having a second analyst independently code 25% of the transcripts; Cohen’s κ = 0.87 (“almost perfect” agreement; Landis and Koch, 1977) confirmed coding consistency. One researcher then applied this codebook to the full dataset, while the second researcher cross-checked a purposive subset of transcripts to verify consistency in code application. Any remaining disagreements were resolved by revisiting the raw data and clarifying code definitions. This multi-step process, together with an audit trail of coding memos, was designed to enhance credibility and dependability in line with COREQ guidelines. Discrepancies were resolved through adjudicative dialogue before finalizing the codebook. A code–recode procedure, re-coding a random subset 3 weeks later, verified the stability of our analytic framework.

Table 11 presents a hierarchical overview of all 63 codes, 21 subthemes, and the seven overarching themes, along with exemplar in-vivo quotations. For full transparency, Appendix Table C includes a frequency ledger detailing segment and interview counts per code, and Appendix Table D provides the original list of 148 open codes generated during the initial coding process. This rigorous, multi-layered approach provides a rich and credible narrative of how situational cues and moral motivations converge to shape ethical consumption decisions.

The following section weaves together two complementary strands: (i) the voices of participants, presented through carefully curated verbatim excerpts, and (ii) theoretical reflections that situate those voices within the study’s S-O-R framework. We first introduce a concise thematic synopsis for each of the seven analytic themes, then illustrate its texture with one to three emblematic quotations (P01, P02, … P22). Quotations were selected based on representativeness and rhetorical clarity; ellipses indicate minor linguistic smoothing that does not alter meaning.

Immediately after each quotation set, we articulate how the expressed reasoning, affect, or behavioral intent aligns with—or nuances—the quantitative path estimates. Where relevant, we highlight tensions (e.g., price-conscious trade-offs) that complicate the straightforward stimulus–response logic, thereby enriching the explanatory power of the S-O-R model. We maintain analytic transparency by clearly indicating the sub-theme and code from which each excerpt was derived.