Introduction

The deterioration of China’s environment is mainly caused by unsustainable production and consumption patterns (Wang et al., 2021b). Since then, consumers’ awareness of environmental protection has grown stronger, taking into account health and the interests of future generations (Tewari et al., 2022). Continuing consumer concerns about consumption patterns and environmental issues are also putting pressure on governments and marketers (Kabadayi et al., 2015). As an alternative to traditional consumption patterns, green consumerism has been widely discussed by marketers and consumers (Tong et al., 2020). Green consumption is a new type of consumption behaviour that conforms to human health and environmental protection standards for the welfare of the whole society and future generations (Chen et al., 2023). Mason et al. (2022) believed that changing consumer patterns, such as choosing greener products, buying sustainable goods, or even promoting shared use, is a key factor in achieving sustainability. However, studies found that consumers’ green behaviour remains underexplored and in a promising stage that requires more in-depth investigations (Sultan et al., 2020; Vita et al., 2019).

Over the years, the consumption attitudes and ideas of the young generation have changed dramatically (Maichum et al., 2017), and they have been regarded as the improvers of environmental problems (Tewari et al., 2022). Young consumers influence the environment not only through their personal behaviour but also through their consumption demand, which can influence the decision-making of enterprises and governments (Hansmann et al., 2020). Because marketers value young consumers’ consumption potential (Gulzari et al., 2022), their perceptions of green consumption provide policymakers with directions for improvement. In addition, consumers born in Generations Y and Z (born after 1980) account for 70 percent of total consumption; thus, making changes to their consumption habits is crucial for achieving sustainable development. It was reported that most of the young consumers receive good environmental education at school, and the effectiveness of this education needs further confirmation (Akhtar et al., 2022).

In addition, young consumers’ consumption behaviour is inexorably impacted by their external environment, such as the information they receive (Cham et al., 2022; Long et al., 2022), cultural values (Halder et al., 2020) and social norms (SNs) (Cham et al., 2023; Kim and Seock, 2019). Normative beliefs have been found to have a significant behavioural impact (Schultz et al., 2016). However, normative social influence is often underestimated (Nolan et al., 2008; Shao et al., 2022). A SN is a widely recognised code of behaviour that has a certain social influence (Lin and Niu, 2018). Individuals often conform to SNs in their daily activities to fit in and avoid social pressure (Aertsens et al., 2009; Ong et al., 2022). Currently, China actively promotes green consumption, and young people frequently receive green education through schools and mass media. Whether this social pressure and consumption orientation will translate into increased green consumption among young consumers remains an empirical question.

Meanwhile, evidence suggests that cultural values play a crucial role in explaining individuals’ environmental intentions (Halder et al., 2020). Incorporating cultural dimensions into research allows for the analysis of cultural norms and characteristics that transcend national boundaries, thereby enhancing explanatory power (Patterson et al., 2006). In East Asian cultures, the concept of “face” occupies significant space in daily actions and thought (Fam et al., 2023; Li and Su, 2007). This face consciousness (FC) reflects individuals’ personality traits, influencing their purchasing decisions and interpersonal communication (Liu et al., 2021). Some scholars believe that green consumption can improve an individual’s status and image (Sun et al., 2017), thanks to the environmental protection attributes and higher prices associated with green products (de Morais et al., 2021). However, the effect of age on green consumption remains mixed in the existing literature (Das et al., 2018; Tan et al., 2022). As a result, there is a lack of research into the specific cultural influences that shape young consumer behaviour in China.

Furthermore, due to a lack of comprehensive information about green products or the high cost of accessing them, consumers may be hesitant to choose green products (Niedermeier et al., 2021). Consumers’ perceptions of the ease of accessing information significantly influence their willingness to choose eco-friendly options (Niedermeier et al., 2021; Shao et al., 2022). When consumers realise that the decision-making process can be simplified by utilising the characteristics of green products, information about the environmental benefits of consuming green can outweigh the price issue and lead them to purchase eco-friendly goods (Joshi and Rahman, 2015). Environmental knowledge (EK), an integral part of environmental literacy, increases people’s understanding of the environment and leads to more environmentally responsible behaviour (Ramdas and Mohamed, 2014). As consumers gain knowledge about green products, they are more likely to choose them.

The extant literature has employed various theoretical frameworks, such as the theory of planned behaviour (TPB), value-belief-norm theory (VBN), and norm activation model (NAM), to examine consumers’ pro-environmental intentions and behaviours. TPB delves into consumers’ pro-environmental behaviour from the perspective of self-interest, while VBN and NAM are grounded in consumers’ altruistic behaviour (Schwartz, 1977). These theoretical foundations have been widely applied in studies on pro-environmental behaviours among young consumers (Ahmed et al., 2021; Becerra et al., 2023). Moreover, Saphores et al. (2012) posit that pro-environmental behaviour is closely tied to the ethical choices of individuals. Consequently, NAM has often been prioritised in research to investigate pro-environmental behaviour among youth (Si et al., 2022).

However, some scholars have criticised NAM for its predominantly altruistic perspective, contending that it places excessive emphasis on ethical considerations while overlooking external social concerns (Long et al., 2022). Compared to conventional alternatives, consumers often perceive green products as expensive and premium (Chekima et al., 2016), and they tend to prefer them only when they personally feel responsible for environmental well-being (Confente and Scarpi, 2021). Therefore, the act of consuming green products inherently involves moral considerations. Upon reviewing previous studies, it becomes apparent that NAM requires expansion rather than simple application to various scenarios. As argued by Han et al. (2019), effectively interpreting consumers’ pro-environmental intentions and behaviours necessitates broadening the scope of the NAM.

Given these research gaps, this study aims to integrate FC, information availability (IA), SNs, and consumers’ EK into the NAM to investigate young Chinese consumers’ green consumption intentions. Particularly, we aim (1) to expand the NAM by including cultural and external environmental factors, (2) to assess the relative significance of FC in young consumers’ green consumption intention in China, (3) to examine the IA’s influence on the model in building intentions, and (4) to investigate how EK and SNs influence young consumer purchasing intention.

Theoretical framework and hypothesis development

Norm activation model

NAM was proposed by Schwartz (1973) and is widely used to explain and predict a range of altruistic and sustainable behaviours. Its applications extend beyond individual consumers to encompass diverse contexts, including manufacturing industry adoption of green IT (Asadi et al., 2019), tourists’ environmentally responsible behaviour (Wu et al., 2022), and restaurant food waste mitigation (Long et al., 2022). NAM believes that people will act altruistically for the benefit of society and the environment, even if these actions are sometimes against their self-interest (Ding Li et al., 2019). According to the NAM model, altruistic behaviour is a noble manifestation of value, and people will internalise social responsibility and moral obligation and embody this value as personal norms. While societal expectations often aim to foster an environment of helping others, individual differences ensure that not everyone readily conforms to such a social orientation. This inherent heterogeneity translates into variations in the generation of altruistic behaviour, with personal norms acting as key triggers (Munerah et al., 2021). The fundamental construct of NAM is Personal Norm (PN), which also has two activators: ascription of responsibility (AR) and awareness of consequence (AC) (Onwezen et al., 2014; Wu et al., 2022).

PN is defined as individuals’ moral expectations, which can inspire an individual’s inner sense of responsibility (Pearce et al., 2022) and are closely related to consumers’ intrinsic values and moral obligations (Rezaei et al., 2019; Ru et al., 2019). For example, when a consumer’s behaviour aligns with their personal norms, they tend to experience positive emotions like pride, self-respect, and security (Kutaula et al., 2022). Consistent with Schwartz (1977), PN is deeply rooted in the self-expectations and obligations arising from an individual’s internalised values. Prior studies have established a positive link between PN and intentions. For example, Choi et al. (2015) showed that PN effectively explained consumers’ intentions to visit a green hotel. In a similar vein, Pearce et al. (2022) found that individuals holding positive PNs are more likely to engage in pro-environmental behaviours.

AR refers to the feeling of responsibility for the problems (Steg and De Groot, 2010). Past studies consistently reported that when people perceive that their actions might lead to negative consequences, their sense of responsibility increases (Schwartz, 1977; Tan et al., 2019). This aligns with Ebreo et al.’s (2003) study, which found that individuals tend to engage in waste reduction behaviour when they feel personally responsible for waste generation. A favourable effect of AR on PN has been demonstrated in prior research (Munerah et al., 2021; Wu et al., 2022). For instance, Vaske et al. (2015) found that increasing AR boosts consumers’ PN in a study on carbon footprint reduction. Similarly, Wu et al. (2022) investigated tourists’ environmentally responsible behaviour and found that AR positively influences PN.

AC, in contrast to AR, focuses on how individuals’ actions might negatively impact others (Gkargkavouzi et al., 2019). NAM posits that people are motivated to act altruistically for the benefit of society or the environment, even if these actions are sometimes against their self-interest (Li et al., 2019). Therefore, from a moral standpoint, people are more likely to act sustainably when they realise their behaviour has adverse environmental consequences (Klockner and Ohms, 2009). Empirical evidence supports this, with Liu et al. (2017) demonstrating that AC positively influences PN, encouraging people to adopt environmentally friendly modes of transportation. Similarly, D’Arco and Marino (2022) found a positive and significant association between AC and environmental citizenship behaviour, mediated by PN in both the private and public spheres.

H1: PN positively and significantly influence youth consumers’ green consumption intention.

H2: AR positively and significantly influence youth consumers’ PN.

H2b: Youth consumers’ PN mediate the impact of AR on buying green products positively.

H3: AC positively and significantly influence PN.

H3b: Youth consumers’ PN mediate the impact of AC on buying green products positively.

Face consciousness

FC refers to “individuals’ desire to enhance, maintain, and avoid losing face in social activities” (Qi and Ploeger, 2021). Values are persistent beliefs people form about specific patterns of behaviour. Understanding the diverse characteristics of consumers’ pro-environment behaviour requires recognising the significant influence of cultural values and SNs across different countries (Diamantopoulos et al., 2003). Influenced by Confucianism, Chinese people closely associate face with their social status or reputation, leading them to focus more on the symbolic value of goods or brands (Ding et al., 2022). To enhance their status and project a “face-saving” image, consumers may prioritise buying products with high public recognition to elicit praise from others (Cham and Easvaralingam, 2012; Liu et al., 2021).

Studies have revealed that the moral dimension of FC can prompt consumers towards socially conscious behaviour and preferences for ethical or eco-friendly products. For instance, Ding et al. (2022) found a significant influence of FC significantly influences on consumers’ intention to purchase traceable seafood. Similarly, Qi and Ploeger (2021) observed a positive relationship between FC and consumers’ intention to buy green food. Moreover, Wu et al. (2022) discovered that FC encourages tourists in the West Lake scenic area to adopt environmentally friendly travel practices.

H4: FC positively and significantly influence youth consumers’ green consumption intention.

Social norms

SN refers to the shared expectations and behaviours that guide individuals within a community (Munerah et al., 2021). However, some scholars criticise NAM for neglecting the influence of these external social factors (Long et al., 2022). Fortunately, integrating SN into NAM models has been shown to significantly enhance its explanatory power. Notably, Hunecke et al. (2001) successfully expanded NAM when studying consumers’ travel mode choices. Their research found that SN positively influenced PN generation and the desire to use public transport. Building upon the NAM framework, Han et al. (2019) integrated SN and emotions to explore the decision-making process of consumers choosing pro-environmental green cruise projects when travelling. Their study found that SN inspires customers to actively engage in positive word-of-mouth activities. This, in turn, enhances individual social pressure and increases the likelihood of their participation in the green cruise project. Further expanding the NAM model, Long et al. (2022) incorporated SN alongside other factors to effectively predict young Chinese consumers’ food waste behaviour. This research demonstrates the model’s potential for understanding diverse pro-environmental actions.

Existing research has consistently highlighted the direct influence of SN as a motivational factor on PN (Onwezen et al., 2013). In essence, our behaviours are mainly guided by PN, which is shaped by SN within our social circles (Gleim and Lawson, 2014; Gleim et al., 2013). This close link is further supported by Han et al.’s. (2019) research study on pro-environmental cruise choices, where SN acted as a key driver for positive word-of-mouth behaviour, ultimately influencing individual participation. Further, expanding on this dynamic, Munerah et al. (2021) investigated the green beauty product purchasing habits of Malaysian consumers. Their findings confirmed the crucial role of PN as a link between SN and consumers’ purchase intentions. This suggests that individuals are more likely to engage in pro-environmental behaviours when they perceive these actions as aligning with the SNs around them.

Additionally, numerous prior studies have documented a strong correlation between SN and intention. For example, Youn et al. (2020) found that consumers’ intentions to dine at traditional restaurants were favourably impacted by SN. Similarly, Yeh et al. (2021) observed a significant influence of SN on consumers’ intentions to choose green hotels. The impact of SN on consumers’ intentions to buy vegan products was also validated by D’Souza (2022). Based on the discussion, the following hypotheses are developed:

H5a: SN positively and significantly influence youth consumers’ PN.

H5b: SN positively and significantly influence youth consumers’ green consumption intention.

H5c: Youth consumers’ PN mediate the impact of SN on buying green products positively.

Environmental knowledge

EK encompasses an individual’s capacity to recognise environmental concepts, comprehend environmental problems, and adopt solutions to address them (Fryxell and Lo, 2003). However, some researchers argue that NAM neglects the role of consumers’ own abilities, such as their perceived consumption validity and knowledge base (Munerah et al., 2021; Onwezen et al., 2014). Furthermore, existing research lacks empirical testing of whether knowledge effectively triggers the specification activation process, a key component of NAM (Ünal et al., 2018). Koo and Chung (2014) proposed that mastering EK is a prerequisite for engaging in pro-environmental behaviour, which itself can serve as a reflection of EK. Individuals with greater EK are more likely to exhibit pro-environment behaviour (Vicente-Molina et al., 2013). However, Kennedy et al. (2009) found a contrasting perspective among Canadian respondents, where a lack of EK emerged as the primary barrier to adopting sustainable practices.

The link between EK and pro-environmental intentions or behaviours is widely acknowledged by researchers. For example, Sharma and Foropon (2019) found that there is a direct association between EK and green purchase intention. Similarly, Wang et al. (2014) identified EK as a key factor explaining sustainable consumption intentions. Sun et al. (2019) observed a direct influence of EK on green product consumption in China. Furthermore, Zameer and Yasmeen (2022) highlighted that knowledge about green products significantly enhances individuals’ desire to make green purchases. In light of the above-discussed evidence, the following hypothesis is developed:

H6: EK positively and significantly influence youth consumers’ green consumption intention.

Information availability

IA refers to the process by which consumers actively seek and gather information about products or services before making a purchase decision (Kumar and Yadav, 2021). In the context of understanding how information influences consumption intention, Momsen and Ohndorf (2022) proposed that readily available product information acts as signals of varying value. This suggests that consumers encounter information that may or may not be helpful in their search for green products. Wang et al. (2021a) further argued that providing consumers with reliable energy-saving information and government-approved certification marks can effectively reduce information barriers. This empowers consumers to make informed choices and manufacturers to prioritise transparency, ultimately leading to the selection of appropriate energy-saving products. Lin and Niu (2018) proposed that readily available environmental information empowers consumers to translate their EK into impactful behaviours. This engagement stems from the utilitarian attitude cultivated by understanding the nutritional value of green food. Therefore, Qi and Ploeger (2021) advocated for marketers to prioritise enhancing nutritional information clarity, accessibility, and efficacy communication for green food products. This strategy aims to cultivate positive consumer perception and ultimately drive informed purchasing decisions.

In addition, IA availability fuels green consumption intentions because green product buyers often seek a deeper understanding of how their choices impact the environment. Maniatis (2016) found that manufacturers advertising their products’ environmental benefits saw a surge in sales, highlighting consumers’ interest in such information. The inconvenience and time investment of independent research can dampen purchase intentions. Therefore, utilitarian shoppers use readily available information to evaluate potential purchases (Kumar and Yadav, 2021; Nystrand and Olsen, 2020). Wu et al. (2021) further confirmed this, demonstrating that positive traceable information about organic food enhances purchase desire. Ultimately, the more information available, the greater the perceived control consumers exert over their purchasing decisions (Khare and Rakesh, 2011). Consequently, it can be hypothesised that:

H7: IA positively and significantly influence youth consumers’ green consumption intention.

The conceptual model constructed in the present study is shown in Fig. 1.

Fig. 1
figure 1

Conceptual framework.

Full size image

Methodology

Measures

To ensure validity and reliability, the survey items were drawn from and adapted from established research. Tewari et al. (2022) provided four items for measuring green consumption intention, while Long et al.’s (2022) four items were adopted for PN. Wu et al.’s (2022) scale was utilised for AC, and Yue et al.’s (2020) established scale was used to measure AR. Inspired by Kumar and Yadav (2021), five items describing consumers’ grasp of green product knowledge were used to measure EK. SNs were assessed using five items, adapted from Munerah et al. (2021). Finally, FC was measured using three items developed by Liu et al. (2021). A two-item modified scale by Kumar and Yadav (2021) was used to measure IA. All items employed a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree”.

A structured questionnaire survey gathered data on both consumers’ green consumption intentions and basic demographic information. Prior to the formal investigation, a pilot study with 35 English-speaking participants was conducted to assess the reliability of the measurement items. Subsequently, the questions were translated into Chinese. Finally, native speakers used back translation on all measurements to ensure accurate content and meaning, following the approach outlined by De Silva et al. (2021).

Data collection

The researchers distributed questionnaires through Wenjuanxing, China’s most popular online survey platform. This platform boasts a nationwide user base, ensuring the representativeness of our sample. The target population comprises Generation Y and Z consumers, recognised as key drivers of green consumption. Therefore, potential respondents were pre-screened, and only those aged 18–43 were invited to participate. During the formal survey, respondents were informed of the survey’s purpose and the estimated completion time. The survey ran from March 13–27, yielding 403 responses. After removing outliers and blatantly illogical submissions, 366 responses remained for analysis. Based on the calculation of G*Power 3.1.9.7, the study had a satisfactory sample size, as it exceeded the minimum sample requirement of 160. In empirical studies related to pro-environmental consumption, it has been indicated that a sample size between 100 and 500 can produce valid and reliable results (Ding et al., 2022; Liu et al., 2021). Table 1 presents the participants’ demographic information.

Table 1 Demographic profile.
Full size table

Data analysis and result

A hybrid PLS-SEM-ANN approach was employed for statistical analysis in the present study. Initially, the relationship between latent constructs was tested using PLS-SEM (Hair et al., 2019), with Smart-PLS 4 used to assess the inner and outer models (Ringle et al., 2015). However, PLS-SEM has limitations in estimating nonlinear relationships between variables (Wang et al., 2022). While artificial neural networking (ANN) analysis can explore nonlinear relationships, it lacks hypothesis testing. This limitation necessitated a hybrid approach, combining ANN with PLS-SEM to unlock a more robust and predictive research model (Leong et al., 2020). The steps of data analysis are illustrated in Fig. 2.

Fig. 2
figure 2

Data analysis steps.

Full size image

Common method variance

Responding to the same questionnaire can introduce unavoidable systematic errors, potentially leading to common method variance (CMV) (Sharma et al., 2021). CMV can diminish the validity of data and hinder hypothesis testing (Hair et al., 2019). Hence, this study employed two methods to assess CMV. Firstly, the Variance Inflation Factor (VIF) (Kock and Lynn, 2012) was utilised, a common tool for detecting collinearity issues (Kumar and Yadav, 2021). When the VIF value is less than 3.3, it is considered that CMV does not exist. Our analysis revealed that all measured items had VIF values below 2.8, suggesting that CMV is not a concern in this study. Additionally, a “method” approach was employed to estimate potential CMV issues (Liang et al., 2007; Low et al., 2023). As shown in Table 2, the average Rs2 to Rm2 ratio of 98.25 further confirms the absence of CMV in this study.

Table 2 CMV analysis.
Full size table

Measurement model assessment

All measurement items in this study exhibited Cronbach’s alpha values exceeding 0.7, as shown in Table 3. This suggests that the data possesses satisfactory internal consistency. The values of standardised factor loadings (>0.7) and composite reliability (>0.7) surpass the thresholds recommended by Hair et al. (2014). Furthermore, the average variance extracted (AVE) affirms the convergent validity of all items, exceeding the cut-off value of 0.5.

Table 3 Construct validity and reliability.
Full size table

In this study, the Heterotrait–Monotrait ratio of correlations (HTMT) was employed to assess the discriminant validity of constructs. HTMT reflects the ratio of the average correlation between items within the same construct (Cheah et al., 2018; Henseler et al., 2015). While a value exceeding 0.9 indicates poor discriminant validity (Henseler et al., 2015), Hair et al. (2019) proposed a stricter criterion of HTMT being below 0.85. As presented in Table 4, the HTMT values for this study demonstrate good discriminant validity.

Table 4 HTMT ratios.
Full size table

Structural model-hypothesis testing

This study employed SmartPLS 4 bootstrapping analysis with 5,000 randomly selected sub-samples to test the hypotheses of the structural model (Hair et al., 2021). The results, presented in Table 5 and Fig. 3, indicate that PN, SN, EK, and IA are significantly and positively linked with GCI, AR, and AC, and SN has a significant positive impact on PN. Therefore, H1, H2, H3, H5a, H5b, H6, and H7 are accepted, while H4 is rejected. Effect sizes (f2) were applied to assess the predictive relevance of research hypotheses, with the f2 values presented in Table 5 ranging between 0.015 and 0.320. These values indicate that the research model’s hypotheses have small to medium effect sizes.

Table 5 Assessment of the structural model.
Full size table

Mediation effects of personal norm

This study posits that subjective norms (SN), ascription responsibility (AR), and awareness of consequences (AC) indirectly influence young consumers’ green consumption intentions through personal norms (PN). The mediation effect is evaluated using the bias-corrected and accelerated bootstrapping method, as indicated in Table 6. The indirect effects of subjective norms, ascription responsibility, and awareness of consequences on green consumption intention are significant, supporting H2b, H3b, and H5c. Following the suggestion of Hair et al. (2021), we conclude that PN partially mediates the relationship between SN and GCI. This conclusion is drawn based on SN having both a significant direct effect (β = 0.163, p < 0.05) and an indirect effect (β = 0.172, p < 0.001) on GCI.

Table 6 Results of mediation analysis.
Full size table

Predictive relevance and PLS Predict

As displayed in Table 7, the value of R2 shows that, on average, 60.2% of the variance in GCI and 61.7% of the variance in PN can be explained by the variation of the independent variables in the model (Shmueli et al., 2016). In terms of Q2, the values of PN and GCI are obviously above zero, demonstrating the model has predictive relevance (Hair et al., 2017).

Table 7 Results of R2 and Q2.
Full size table

PLS Predict procedure with 10 folds was conducted, following Hair et al. (2019). As shown in Table 8, GCI’s and PN’s Q2 prediction values are greater than zero, indicating that the model’s predictive power is reliable. In addition, the RMSE values of constructs calculated by the PLS-SEM method are all smaller than those calculated by LM (except PN1), suggesting the considerable predictive power of this study (Munerah et al., 2021).

Table 8 PLS predict assessment.
Full size table

Importance performance map analysis

Importance performance map analysis (IPMA) enables the direct assessment of construct performance by revealing their average latent variable scores (Wang et al., 2019). Therefore, IPMA visualises the impact of each construct on the dependent variable through graphs. Figure 4 displays the IPMA results for GCI. Among the constructs, PN (0.470) exhibits the best performance, followed by AC (0.365) and AR (0.330). Nevertheless, FC (0.006) falls slightly below the average in importance.

Fig. 3: Structural model test results.
figure 3

Notes: *p < 0.05; **p < 0.01; ***p < 0.001.

Full size image

Artificial neural networking analysis

ANN addresses the limitation of PLS-SEM, which solely studies linear hypothesis relations and has gained widespread application in decision-making tasks (Wang et al., 2022). In this study, ANN was employed to assess the robustness of PLS-SEM and explore latent variables. Both the GCI and PN models in this study reveal significant variables. SPSS’s artificial neural network analysis was used to establish models for GCI, PN, and their respective significant variables. The Root Mean Square Error (RMSE) serves as a measure of the error between the training and test data. As shown in Table 9, the average difference between the training and test RMSE values in both models is less than 0.5 (Sharma et al., 2021), indicating that these models have predictive accuracy and effectively capture the relationships between significant variables and predictors.

Table 9 ANN- RMSE.
Full size table

The employed ANN model utilises a three-tiered layer structure (Leong et al., 2020). The sigmoid function was chosen as the activation function for its advantages in handling both low- and high-end data (Sharma et al., 2021). In this study, PN, SN, EK, and IA constitute the input layer, while GCI forms the output layer. Notably, AR, AC, and SN serve as the input layers for PN. This structure is depicted in Figs. 5–6. The ANN model effectively reveals the nonlinear relationships between variables. Furthermore, the combination of PLS-SEM and ANN enhances the model’s explanatory power (Sharma et al., 2021).

Fig. 4
figure 4

IPMA for GCI.

Full size image
Fig. 5
figure 5

Neural network architecture for GCI.

Full size image
Fig. 6
figure 6

Neural network architecture for PN.

Full size image

Sensitivity analysis provides an intuitive ranking of significant variables (Mishra et al. 2022), calculated by dividing the average importance of each significant variable by its largest importance. The results are presented as percentages within each ANN model (Sharma et al., 2021). Table 10 reveals that PN has the strongest predictive ability for GCI, which aligns with the PLS-SEM results. However, the order of influence for other significant variables differs between the two methods. Additionally, AR emerges as the strongest predictor for PN, followed by AC and SN, mirroring the PLS-SEM findings.

Table 10 Sensitivity analysis.
Full size table

Discussion

While several studies have explored the green consumption intentions of younger generations, few have specifically examined the influence of “face”, a key concept in traditional Chinese culture, on young consumers’ green choices. To address this gap, this study extends the NAM model by incorporating EK, IA, subjective norms, and FC to predict green consumption intention in young consumers. This study employed a two-stage PLS-SEM-ANN approach to analyse the collected data. The findings show that, with the exception of FC, all other variables significantly impact green consumption intention. Additionally, the ANN results suggest that further investigation is necessary to fully comprehend the influence of other external factors.

Our results provide robust support for H1, demonstrating that PN significantly enhances young consumers’ GCI. This finding aligns with Gleim and Lawson (2014) and Rezaei et al. (2019), who concluded that personal norms ultimately determine whether individual decisions translate into action. As posited by Schwartz (1977), personal norms represent our internal moral convictions. In the context of green consumption, which directly addresses environmental concerns, a strong moral dimension is crucial (Hunecke et al., 2001). Young consumers, having grown up witnessing China’s environmental pollution first-hand (Lee, 2008), possess heightened sensitivity to environmental issues, making them more likely to form internal moral constraints that drive green consumption behaviour. In addition, personal norms are influenced by SN, reflecting people’s tendency to conform to group expectations (Song et al., 2019). This explains the significant influence of SN on PN, supporting H5. This finding is in line with Thøgersen’s (2006) assertion that subjective norms play a direct role in shaping personal norms, especially when analysing sustainable environmental behaviour.

AR and AC demonstrate significant positive relationships with PN, supporting H2a and H3a. This finding aligns with established research by Han et al. (2021) and Munerah et al. (2021). Furthermore, the mediating role of PN within the relationships between AR, AC, SN, and green consumption intention is confirmed, offering support for H2b, H3b, and H5c. Notably, AR and AC are foundational variables in the NAM model, indirectly influencing intention through their impact on PN (Schwartz, 1977). This implies that a stronger perception of environmental consequences and a greater sense of personal responsibility led to more robust personal norms, ultimately driving green consumption behaviour (Song et al., 2019).

However, contrary to H4, this study found that FC does not significantly influence young consumers’ GCI. This finding departs from previous studies that have attributed FC to consumers’ pursuit of conspicuous goods or luxury goods (Islam et al., 2022; Kumar et al., 2021). These studies suggest that consumers opt for conspicuous goods to enhance their social standing and gain face within their social circles. Such purchases, therefore, serve primarily self-interested motives. However, the primary objective of purchasing green goods is to contribute to environmental sustainability, a goal driven by altruism and with a higher moral compass among consumers. Previous studies have established a strong relationship between income and luxury consumption (Dubois and Duquesne, 1993). Income might be a more influential factor in deterring young consumers from choosing green products over luxury goods, rather than “face” concerns.

This study also found that EK significantly promotes young consumers’ desire to buy green products, thereby confirming H6. This finding aligns with research suggesting that consumer knowledge directly translates into behaviour (Hosta and Zabkar, 2021). The environmental education young consumers receive has significantly enhanced their EK and awareness (Zsóka et al., 2013). Greater EK empowers consumers to recognise the environmental implications of production and consumption practices when making purchasing decisions, consequently influencing their choices (Lin and Niu, 2018; Roh et al., 2022).

The next point that needs to be discussed is the positive relationship between IA and GCI, which is consistent with the findings in Kumar and Yadav (2021) and Hosta and Zabkar (2021), thus supporting H7. In sustainable consumption, readily available information plays a crucial role. Limited information can hinder consumers’ desire to purchase green products (Laato et al., 2020). Fortunately, young consumers have access to real-time information about green products through various online platforms. This convenient access not only saves them money but also allows them to stay informed about environmental issues. Information released by the government and enterprises about the environmental consequences of unsustainable consumption can further heighten their environmental sensitivity.

Conclusion

Theoretically, this study expands the boundaries of green consumption research by incorporating FC and IA into the NAM framework to analyse GCI in young Chinese consumers. While the existing literature has explored FC’s role in luxury or conspicuous goods consumption (e.g., Islam et al., 2022; Zhang and Wang, 2019), this study extends this concept to the green consumption literature. Recognising the additional influence of IA on young consumers’ decisions, this study offers new insights for producers and policymakers. Furthermore, integrating concepts like EK and SN into the green consumption study addresses the limitations of the NAM model and provides a valuable theoretical foundation for future research. In addition to the above, the findings also highlight EK as a key antecedent of young consumers’ green consumption. This underscores the need for schools and governments to prioritise environmental education initiatives across social groups. Additionally, the research suggests valuable avenues for enterprise marketing. Highlighting green products’ pro-social image and environmental attributes can resonate strongly with young consumers.

Limitations and future directions

This study suggests several areas for future exploration. First, the primary focus is on the influence of “face”, a traditional Chinese cultural factor, on consumer behaviour. However, East Asian and Western cultures differ significantly. Future studies should use diverse datasets from China to validate the model’s generalisability across different cultural contexts. Secondly, due to budgetary constraints, non-probability sampling was employed in this research, limiting the generalisability of the findings. To enhance the external validity of research results and reduce selection bias, it is recommended that future studies adopt probability sampling techniques to obtain a wider range of data. For instance, we should sample from every province in China to prevent sample clustering. Furthermore, the synergistic effect of antecedents was not demonstrated in this study, and subsequent studies are recommended to use fsQCA to analyse the combined effect of antecedents on outcomes. Thirdly, price sensitivity is known to impact consumer behaviour (Barman et al., 2022) and may have a direct bearing on young consumers’ ability to afford green products. In subsequent studies, we aim to incorporate price sensitivity into the model or investigate its moderating effect.