Sustainable consumption practices among Chinese youth

The paper addresses the growing global concerns about climate change, environmental degradation, and pollution, and positions green consumption as a key pathway to sustainable development aligned with the UN 2030 Agenda. Although consumer interest in green products is rising, actual market shares remain low, particularly among younger consumers, highlighting an intention–behavior gap. Focusing on Generation Y (ages 18–35) in China—who are technologically adept and environmentally concerned but face purchasing power constraints—the study seeks to identify predictors of green consumption behavior. The research question is: What are the predictors of green consumption behavior among Chinese Generation Y consumers? The authors note limitations of the theory of planned behavior (TPB) in fully explaining green behaviors and propose integrating it with the knowledge–attitude–practice (KAP) model, and further differentiating attitudes into three dimensions (towards environmental issues, eco-social benefits, and green consumption). They also examine green self-identity and eco-labeling as moderators of the intention–behavior link to address the green gap.

Theoretical underpinning integrates TPB and KAP. TPB posits that attitude, perceived behavioral control (PBC), and subjective norms determine intentions, which predict behavior, yet prior work shows mixed or insufficient explanatory power for green behaviors. KAP, widely used across domains, examines knowledge, attitudes, and practices, and helps identify gaps and barriers. Prior studies indicate environmental knowledge and concern influence attitudes and green purchasing, suggesting value in extending TPB with these constructs. The paper refines attitudes into three facets: attitudes towards environmental issues, eco-social benefits, and green consumption, to capture nuanced pathways from knowledge/concern to intention. It also addresses the attitude–intention and intention–behavior gaps by proposing green self-identity and eco-labeling as moderators of the intention–behavior relationship. Although eco-labels often influence awareness and trust, their moderating role between intention and behavior has been underexplored; similarly, green self-identity has been used as a moderator for other links but not commonly for intention–behavior in this context. The study thus proposes and tests a comprehensive KAP–TPB framework with additional constructs and moderators, and explores subgroup differences by gender and income using multigroup analysis.

Design: Positivist, quantitative, cross-sectional survey integrating KAP and TPB. Eleven hypothesized relationships were tested using PLS-SEM (Smart-PLS 4.0). Sampling and data collection: Convenience sampling of Chinese youth (18–35) via online platform WJX and dissemination on WeChat; fielded April–June 2023. Of 902 responses, 876 valid cases remained after removing 26 straight-liners. Sample characteristics: Gender roughly balanced (55.8% female), ages spread across 18–35, diverse education and income; most reported sometimes or rarely consuming green products; majority spent less than RMB 1000 per month on green consumption. Instruments: Two sections—A: demographics; B: constructs measured with 5 items each on 7-point Likert scales. Constructs: environmental concern, environmental knowledge, attitudes towards environmental issues, eco-social benefits, and green consumption, subjective norms, PBC, eco-labeling, green self-identity (GS), green consumption intention (GCI), and green consumption behavior (GCB). Items sourced/adapted from prior literature; all items listed in Supplementary Material S1. Analysis: Two-stage PLS-SEM analysis (measurement then structural) with mediation and moderation testing. Common method bias assessed via Harman’s single-factor test (single factor = 37.7% variance) and full collinearity VIFs (1.253–2.025), indicating minimal bias. Data were non-normal (Mardia’s tests), supporting PLS-SEM use. Reliability and validity were satisfactory: Cronbach’s alpha and composite reliabilities > 0.90 across constructs; AVE ~0.722–0.759; VIFs below 2.5; Fornell–Larcker and HTMT (<0.90) supported discriminant validity. Predictive relevance assessed via PLS-predict (Q² > 0; PLS RMSE generally lower than linear model baseline). Multigroup analysis (MICOM) used to assess invariance and differences by gender and income.

Measurement model: All constructs exhibited strong reliability (Cronbach’s alpha and composite reliability typically >0.90) and convergent validity (AVE ~0.72–0.76). Discriminant validity met via Fornell–Larcker and HTMT (<0.90). CMB not critical (single factor 37.7%; full collinearity VIFs 1.253–2.025). Structural model highlights (betas, t-values, p-values): - Determinants of attitudes: • Environmental concern → Attitude towards environmental issues: β=0.263, t=6.531, p<0.001 (accepted). • Environmental knowledge → Attitude towards environmental issues: β=0.330, t=8.309, p<0.001 (accepted). • Environmental concern → Attitude towards eco-social benefits: β=0.221, t=5.259, p<0.001 (accepted). • Environmental knowledge → Attitude towards eco-social benefits: β=0.344, t=8.068, p<0.001 (accepted). • Environmental concern → Attitude towards green consumption: β=0.330, t=8.205, p<0.001 (accepted). • Environmental knowledge → Attitude towards green consumption: β=0.312, t=7.465, p<0.001 (accepted). - Determinants of intention: • Attitude towards environmental issues → GCI: β=0.038, t=0.771, p=0.221 (rejected). • Attitude towards eco-social benefits → GCI: β=0.089, t=1.671, p=0.047 (accepted, small effect). • Attitude towards green consumption → GCI: β=0.097, t=1.959, p=0.025 (accepted, small effect). • Subjective norms → GCI: β=0.206, t=4.315, p<0.001 (accepted). • Perceived behavioral control → GCI: β=0.074, t=1.582, p=0.057 (rejected). - Determinants of behavior: • GCI → GCB: β=0.090, t=2.848, p=0.002 (accepted). • PBC → GCB: β=0.236, t=5.534, p<0.001 (accepted). • Eco-labeling → GCB: β=0.207, t=5.428, p<0.001 (direct positive relation reported in structural figure/text). - Moderation: • Eco-labeling × GCI → GCB: β=−0.055, t=1.555, p=0.060 (not significant; rejected). • Green self-identity × GCI → GCB: β=0.272, t=6.487, p<0.001 (significant; accepted). Model performance: R² for GCB=0.392, Attitude towards green consumption=0.307, Attitude towards environmental issues=0.261; Attitude towards eco-social benefits=0.240; GCI=0.157. Effect sizes were generally small (f² from 0.004 to 0.120). PLS-predict showed Q²>0 and lower RMSE than LM for most indicators. Multigroup analysis: MICOM suggested mostly invariant measurement across gender and income groups; no significant differences were found across most hypothesized paths by gender or income, with some nuanced variations noted (e.g., potential differences in GCI→GCB by income in MICOM permutation), but overall no significant invariance violations.

Findings show that both environmental concern and knowledge significantly and positively shape all three attitudinal dimensions, confirming the KAP premise that knowledge and concern foster pro-environmental attitudes. Among attitude facets, attitudes towards eco-social benefits and towards green consumption translate into intention, while attitudes towards environmental issues do not—suggesting that pragmatic and socially oriented evaluations matter more for intention than general environmental issue attitudes. Subjective norms play a meaningful role in intention, indicating social influence as a lever for youth green intentions. Perceived behavioral control does not significantly predict intention but does directly predict behavior, implying that control perceptions may activate behavior more than they motivate intentions in this context. Intention predicts behavior, but with a small effect, reflecting the well-documented intention–behavior gap. Green self-identity strengthens the intention–behavior link, highlighting the importance of identity alignment in converting intentions into action, whereas eco-labeling did not moderate this link. Overall, the integrated KAP–TPB model (with multidimensional attitudes and identity moderation) provides a nuanced account of green consumption behavior among Chinese youth and indicates that education, social norms, and identity-based interventions may be more effective than solely raising general environmental attitudes.

The study advances understanding of Chinese youths’ green consumption by integrating KAP and TPB, decomposing attitudes into three dimensions, and testing moderators of the intention–behavior relationship. Key contributions include: (1) validating that environmental concern and knowledge feed into attitudes, (2) demonstrating that attitudes toward eco-social benefits and green consumption, along with subjective norms, drive intention, (3) confirming that perceived behavioral control and intention predict behavior, and (4) identifying green self-identity as a significant moderator that strengthens the intention–behavior pathway while eco-labeling does not. The model explains a substantive share of variance in behavior (R²=0.392). Practical directions include enhancing environmental knowledge and concern through education, emphasizing eco-social benefits in campaigns, leveraging social norms and youth communities, and fostering green self-identity to bridge the intention–behavior gap. Future work should further refine moderators (e.g., green trust, eco-literacy) and consider longitudinal designs to track changes over time.

The cross-sectional design offers only a snapshot and cannot capture longer-term behavior change. Reliance on self-reported measures introduces potential response and social desirability biases, especially among youth. An intention–behavior gap persists, with some hypothesized intention-related effects not translating into significant behavior predictors. The multigroup analysis considered only gender and income; other demographic or contextual factors (e.g., region, education level, household composition) may yield additional insights. Future research should employ longitudinal or experimental designs, incorporate objective behavioral data where possible, and test additional moderators such as green trust and eco-literacy to better address attitude–intention and intention–behavior gaps.