This study examines the determinants of green purchase intention (GPI) and its relationship with sustainable consumer behavior, namely green technology adoption (GTA) and eco-friendly product repurchases (EC). The paper also explores the moderating role of green self-identity (GSI) in these relationships. Using the quantitative PLS-SEM method, the research results show that GPI plays an important mediating role in connecting environmental factors with green consumption behavior. Additionally, the authors have explored the moderating role of GSI in both GPI-GTA and GPI-EC relationships. The study contributes to theory by expanding the framework of green consumer behaviour by including self-identity, a meaningful psychosocial construct in the context of sustainable consumption. Furthermore, the research results provide a basis for policymakers and businesses to develop communication, education, and promotion programs to encourage green consumer behavior, especially in the selection, use, and purchase of environmentally friendly products. This shows the need to raise awareness of environmental protection, promote people's green consumption intentions and spread the image of environmentally responsible consumers in the community.

A quantitative research method was used by utilizing Partial Least Squares Structural Equation Modeling (PLS-SEM) in SmartPLS 4 to analyze the relationships between latent variables. The questionnaire responses were gathered by means of a structured questionnaire with the response scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree). The items for measurement were taken from some valid scales as follows:

First, EA, PBC, and GPI are adopted from TPB to align with the green consumption context in Vietnam and slightly modified.

Second, GPV with some adjustments to address the specific consumer concerns in the Vietnamese market, based on the study by Chen & Chang (2012).

Third, based on Haws et al. (2014) and Zhuang et al. (2021), the environmental knowledge (EK) scale was modified to align with Vietnamese consumers' environmental awareness.

Finally, the Green Self-Identity (GSI) was slightly adjusted to reflect the social characteristics of green consumers in Vietnam, based on the study of Confente et al. (2020).

The questionnaire was initially developed in English and then translated into Vietnamese using a forward–backward translation procedure to ensure semantic and conceptual equivalence. A pilot survey was then implemented with 50 consumers to assess clarity, linguistic accuracy, and cultural appropriateness in Ho Chi Minh City - Vietnam, after which minor revisions were corrected, which aims to improve the final version.

Reliability was assessed using Cronbach’s alpha and Composite Reliability (CR)> 0.70, and convergent validity was established with standardised loadings> 0.70; Average Variance Extracted (AVE) > 0.50 is considered good. Discriminant validity was tested by the Heterotrait–Monotrait ratio (HTMT) (< 0.85/0.90).

Procedural Measures Several remedies were applied to reduce Common Method Bias (CMB), such as ensuring full anonymity, randomizing the item order, using neutral wording, and distinguishing between predictor and criterion constructs. There was no multicollinearity issue, as the maximum VIF was < 3.0, indicating that multicollinearity was not a significant problem.

Univariate (|z| > 3) and multivariate (Mahalanobis, p < 0.001) outliers were removed, and missing data were handled by listwise deletion. PLS-SEM model was estimated using a path-weighting scheme (300 iterations, 1e–7 stopping criterion) and bootstrapping 5,000 samples (two-tailed, α = 0.05). SRMR, NFI, and Chi-square (χ²) were used to assess model fit. Due to the cross-sectional design, results are interpreted as associations rather than causal relationships. R², adjusted R², and f², as indicated by Cohen (2013) are reported to express the extent of explanation and effect.

A cross-sectional design was used because the study focused on measuring awareness, attitudes and behaviors at one point in time. Psychological constructs (GPV, EA, EK, PBC, GPI, GSI) are relatively stable in the short term, so a one-time data collection is sufficient to determine the relationship between variables. At the same time, the green consumption context in Vietnam is changing rapidly, so a point-in-time survey helps to accurately reflect current behavioral reality.