Mechanisms of Media Persuasion and Positive Internet Word-of-Mouth Driving Green Purchasing Behavior: Evidence from China

Abstract

As environmental issues intensify, sustainability development is becoming mainstream, with environmental topics gaining increasing attention in the media and online. Shifting consumer behavior in China toward green purchasing is crucial for mitigating environmental pollution and achieving sustainable, low-carbon consumption. This study constructed a theoretical model combining media persuasion (MP) and positive internet word-of-mouth (PIM) with green purchasing behavior (GPB), based on the Stimulus–Organism–Response (SOR) and persuasion theories, to explore consumer responses to environmental information campaigns. A total of 357 valid samples were collected through an online questionnaire survey and subjected to analysis using the structural equation model (SEM). The results indicate that MP, PIM, and environmental attitude (EA) significantly influence GPB. Specifically, EA partially mediates the relationship between MP, PIM, and GPB, while environmental knowledge (EK) negatively moderates the relationship between independent variables and EA. Additionally, EK moderates the mediating effect of EA. The findings highlight that the effective implementation of MPs and PIMs can facilitate the creation of positive EA, which stimulates consumer GPB. This is essential for promoting sustainable consumption. This research contributes to sustainability by providing insights and practical suggestions for developing green marketing strategies that support environmental goals.

1. Introduction

In recent years, significant developments in industrialization, urbanization, and informatization have occurred. New media elements have been integrated into every aspect of life, drastically changing people’s lifestyles and consumption patterns [1]. Nevertheless, the swift expansion of the economy and heightened levels of consumption have given rise to a range of environmental concerns, including air and water contamination and waste management challenges, which significantly impact people’s quality of life [2]. GPB refers to pro-environment actions crucial for ecological protection, resource conservation, and sustainable development. Encouraging the demand for green products and services is essential to promoting environmental sustainability [3]. As a result, understanding the underlying motivation and driving factors of GPB and promoting it among consumers has become a widely discussed topic among scholars and the community.

Sustainable global development is the order of the day [4]. When consumers make environmentally friendly purchases, they prefer products with minimal resource depletion and high recycling efficiency [5]. GPB effectively reduces environmental impacts and supports sustainable development [6]. Its contribution to the ecological environment is huge through policy guidance, social shaping, and calls for GPB [7]. Companies are actively exploring the development path of green practices [8,9]. In response to the policy, pertinent green goods and services were introduced to satisfy customers’ demands for environmentally friendly items [10].

In China, governments have issued many initiatives to promote green development, such as the dual-carbon strategy proposed in 2020 for macroeconomic control [11]. However, the promotion of green purchasing in the implementation process is unsatisfactory, and it is even more frustrating that the “Plastic Restriction Order” has evolved into the “Plastic Selling Order” [12]. It can be seen that green persuasion through social appeals and the media alone is only a moral exhortation to the public while ignoring actual consumer demands. It cannot enhance the efficiency of environmental governance in the green trade [13].

In the rapidly developing information age of China, it is important to thoroughly explore how media information dissemination can effectively shape the creation of a society focused on green development [14]. Consumers are constantly surrounded by numerous new media channels saturated with information about different products and services. Consumers trust non-interest groups’ evaluations of companies’ products and services (media and internet users’ word-of-mouth) more than product details obtained from advertising and corporations [15]. However, GPB is characterized by rationality, high cost, and altruism [16]. Consumers acquire knowledge regarding green items through widely accessible media channels including the internet and television programs, which aids them in making informed choices when making purchases [17]. Making full use of the persuasive power of the media and word-of-mouth on the internet helps consumers to be orientated toward green purchasing [18].

Furthermore, a survey conducted by China Youth Daily on 15th March 2021 found that 90% of participants referred to user evaluations released by online platforms and social media software. Furthermore, 46% of participants reported that they would be influenced by internet word of mouth (IWOM) when making purchasing decisions [2]. This highlights the significant impact of IWOM and ratings on purchase intentions and behavior. Existing research has delved into the process of GPB and the willingness to purchase based on different theories such as the Theory of Planned Behavior (TPB) [19], Norm Activation Theory (NAT) [20], and Value–Belief–Normative (VBN) theory [21]. While these theoretical models can better predict the causes of consumers’ green purchasing attitudes or behaviors, in practice, it is found that many consumers’ positive environmental attitudes do not translate into actual behaviors. In other words, there is a bias between green attitudes and behaviors [22,23,24]. Hence, disseminating green information is a crucial factor influencing consumers to purchase environmentally friendly products. This holds great significance for developing marketing strategies for enterprises, especially in promoting green products.

This paper will answer the following questions with confidence: Do MP and PIM shape EA and drive GPB? Is there a substantial disparity in the efficacy of MP and PIM in the dissemination of information? Does EK affect their relationship? Hence, this paper confidently discusses issues with the persuasion theory and SOR framework, using Chinese online consumers as the research population. The aim is to confidently provide a practical reference for motivating and guiding GPB.

2. Literature Review

2.1. Underpinning Theory

To better explore the formation process of consumers’ GPB and avoid the bias between green attitudes and behaviors, two important grounded theories are applied in combination. The SOR framework is used as an entry point to consider the new media environment with Chinese characteristics. Two constructs, MP, and PIM, are abstracted based on persuasion theory to explore consumers’ attitudes and reflections when stimulated by external information.

2.1.1. S–O-R

In studies of behaviorist learning theory, Gleitman et al. (1954) [25] suggested that all complex human behaviors can be broken down into two parts: stimulus and response, and that behavior is a response to a stimulus. However, the S-R model fails to consider the organism’s consciousness and the intricate process of complicated mental operations. The SOR model has been proposed to address this problem effectively. The organism’s response to stimuli involves an intermediate cognitive and mental processing stage, in addition to the direct reaction to the input [26]. The theory has been used to predict social media behavior [27], online consumption [28], and GPB [29].

Firstly, the stimulus (S) is the sum of internal and external drivers, physical factors that exist objectively in the real world and do not have a specific range [30,31]. Namkung and Jang (2010) [32] chose the shopping environment of the online platform as a stimulus, which translated into the consumers’ emotions and cognitively driven shopping behavior. Yang and Zhang (2020) [12] used the media communication environment as a stimulus to investigate the mechanisms by which MP influences consumer behavior. This paper considers the effect of PIM and MP on consumer stimulation in the context of the information society.

Second, the organism (O) reflects the mental state of the organism in the face of external stimuli [33]. It manifests as an individual’s emotions and perceptions and encompasses the mental states of pleasantness, arousal, and control [34]. When observing an organism’s mental state, it is important to consider the specific research situation, such as shopping satisfaction [35]. Environmental awareness and status awareness act as intermediaries between consumer confidence and intention, as they are organisms that are subjected to external stimuli [36]. Hence, this paper adopts the EA as the organism to mediate.

Finally, the response (R) represents the stimulated person’s willingness to act and eventual behavior [37]. The results of organismic switching encompass both converging and avoiding behaviors. This entails approaching or moving away from a target determined by the stimulus force [38]. These responses include usage behavior [39] and purchasing intention [40]. A combination of internal and external variables impacts consumers’ purchase decisions. Their attitudes and motivations are stimulated by products and services, leading them to make a purchase decision. After the purchase, consumers evaluate the goods and related channels and manufacturers, completing the entire purchase decision process.

2.1.2. Persuasion Theory

The persuasive messages conveyed by the mass media are crucial to understanding politics and social change. Nowadays, recognizing and understanding the impact of persuasive messages is more urgent and important than ever [41]. Persuasion theory is the appropriate dissemination of information to change the receiver’s attitude; the process includes the communication–attitude–behavior process [42]. Persuasion theory helps society establish communicative relationships between different factors [43]. Persuasive communication positively affects individuals’ environmental intentions, leading to perceived facilitation and behavioral control [44]. Chiu et al. (2020) [45] posited that the combination of persuasion and the TPB can predict the mechanisms that drive the sustainable behavior of residents.

Factors such as the credibility of the source of the communicator and the order in which the information is communicated can have an impact on the ability to persuade [46]. Hence, this research critically analyzes the influence approach of green marketing within the framework of new media, taking into account the reality of China’s information dissemination. Two independent variables, MP and PIM, are assertively constructed from the perspectives of official media channels and netizens’ autonomous communication channels.

2.2. Green Purchasing Behavior (GPB)

Consumers actively choose green products and services that minimize resource depletion and environmental damage while making purchases. This is known as GPB [47,48]. The population’s heightened environmental awareness has led to choices that are more responsive to and aligned with the global strategy of sustainable development, and the requirements of social development are being taken into consideration [49]. Motivating and guiding consumers to practice environmentally friendly purchasing is crucial for protecting the ecology [6,50]. Academics have extensively analyzed the consequences of green purchasing from three main viewpoints: the individual’s perception [51,52], the influence of the social environment [53], and the green attributes and marketing of products [54,55,56]. This extensive analysis demonstrates a profound understanding of and proficiency in the subject and emphasizes the significance of taking into account all variables when making well-informed judgments regarding green purchases.

Alghamdi and Agag (2024) [10] claimed that environmental perceptions, psychological identity, and demographic factors can greatly influence the intents and actions related to GPB. Xu et al. (2022) [57] demonstrated the significant effect of moral emotions and environmental responsibility on GPB, which may be influenced by external economic and social factors; however, it is important to note that this influence might diminish over time [58]. Ultimately, individual buying intentions are influenced by a blend of external and internal variables. Scholars have thoroughly investigated the effectiveness of communication strategies, such as advertisements, and confirmed their significant influence on consumers’ attitudes and perceptions by disseminating relevant environmental information [59,60].

As a result, consumers’ practice of green behaviors is positively impacted. The advancement of information technology has resulted in a change in emphasis in conventional media advertisements. However, merchants tend to describe their products from their perspective, which is a strategy aimed at promoting sales, but may not fully win the trust of consumers [2]. With the emergence of IWOM, consumers now have more channels to learn about product information. Thoroughly examining the capacity to influence media communication in coordination with official and self-media communication is crucial.

2.3. Media Persuasion (MP)

Persuasion theory suggests that the media, as an external stimulus, can alter consumer perceptions and attitudes by activating internal motivations to change individual attitudes and behaviors in the direction intended by the persuader [12]. MP primarily involves consistently exposing environmental issues across different social media platforms. Through specific appeals to audiences, it aims to steer changes in their attitudes and behaviors according to the media’s predetermined direction [61]. Lee (2010) [62] developed a unidimensional scale of MP to study the frequency of environmental information stimuli that audiences receive through different channels; the scale is easy to use and has been cited by many scholars.

External information stimuli improve individuals’ assessment of the accuracy of environmental information. This strengthens the persuasive power of the media to encourage individuals to adopt environmentally relevant behaviors [63]. Empirical evidence confirms the relationship between MP in environmental applications and GPB [64]. The widespread dissemination of environmental information on global warming through the media can increase public awareness of environmental issues, increasing concern about pollution and thus positively influencing individual environmental behavior [65]. According to [12], MP is effective in shaping consumer GPB and is moderated by the perceived severity of the environment. Increased exposure to energy-saving information in the media leads to higher sales of energy-saving products, demonstrating the significant impact of MP on GPB [66]. In addition, Miller (1976) [67] discovered that the more often people encounter persuasive messages, the more their attitude toward the message improves. Hence, the following study hypotheses were formulated:

H1a. 

MP has a positive influence on EA.

H1b. 

MP has a positive influence on GPB.

2.4. Positive Internet Word-of-Mouth (PIM)

The direction of transmission determines whether IWOM is positive or negative [68]. PIM is the result of consumers’ favorable attitudes toward a particular brand and their positive evaluations of its products. Conversely, negative IWOM arises from consumer dissatisfaction and is generated by passing negative information to others [69]. Researchers have extensively studied the effects of negative IWOM and discovered that it can have a detrimental influence [70] or positive impacts [71]. Corporate marketing often leads users to generate PIM, and positive effects can also have a significant impact on consumer decisions in specific contexts [72], even due to negative IWOM [73]. Hence, this study selected PIM as a variable for the research.

PIM is closely related to sustainable consumer behavior [74]. PIM refers to positive comments about a company’s products and services provided by non-stakeholders to individuals and organizations via internet platforms [59]. Current research on PIM in the green behavior field focuses on the willingness to communicate [75], mediating IWOM effects [76], and influencing factors of GPB [77,78]. PIM communication can significantly influence consumers’ purchase intentions and behaviors [79]. PIM regarding environmental information, such as products or brands, is more likely to increase consumer trust and willingness to implement GPBs [16]. The process of influencing consumers’ green behavior through PIM demonstrated the parallel mediating role of individual factors [80]. Hence, the following research hypotheses were formulated:

H2a. 

PIM has a positive influence on EA.

H2b. 

PIM has a positive influence on GPB.

2.5. Environmental Attitude (EA)

Attitudes are evaluations made by individuals to determine their inclination toward a particular situation [81]. Scholars typically defined EA in the early days as a tendency to be environmentally friendly, ultimately manifested through environmental behavior [82]. As the research progressed, it became clear that EA is more influenced by situational emotions and beliefs, which may not necessarily reflect actual actions [83]. EA is the prevailing attitudes and perceptions that individuals hold about the natural environment [84].

Individuals with a positive attitude toward green products are more likely to engage in pro-environmental behavior [85]. Having a positive attitude toward sustainable development increases the likelihood of purchasing green products or services. [86]. EA is endogenously influenced by an individual’s environmental awareness and green self-identity, effectively contributing to their GPBs and intentions [87]. EA is a key factor in measuring GPB [88]. Individual attitudes significantly influence purchase intentions toward sustainable clothing [89]. Consumers’ positive attitudes toward the environment sustain their green behaviors over time [90]. Given the above, the following hypothesis was developed:

H3. 

EA has a positive influence on GPB.

Furthermore, consumer attitudes can lead to different behavioral outcomes due to internal and external stimuli. It is important to present these attitudes objectively, without any subjective evaluations [88]. EA has a mediating effect between multiple factors and GPB, contributing to improved environmental quality and reduced ecological pollution [91]. For example, EA operates as a mediator between environmental awareness and GPB, effectively influencing customers to engage in sustainable consumption practices [92]. Therefore, the following study hypotheses were formulated:

H4a. 

EA mediates the relationship between MP and GPB.

H4b. 

EA mediates the relationship between PIM and GPB.

2.6. Environmental Knowledge (EK)

EK is the general awareness of the natural environment and ecosystem facts, concepts, and relationships [93]. It represents an individual’s understanding of the natural environment and is a natural interpretation of the real world [50,94]. Scholars have recognized EK as a crucial determinant in assessing GPB. It has the potential to shape individuals’ perspectives on sustainable development and can either positively or negatively affect their choices of consumption [95]. Green marketing, ecological awareness, environmental consciousness, environmentally friendly pricing, and eco-friendly goods exert substantial positive impacts on environmentally conscious buying patterns [96].

EK has a significant moderating role (positive moderating/negative moderating) in the relationship between GPB and multiple factors [97]. According to [98], the possession of EK strengthens the influence of perceived environmental value on individuals’ intentions to engage in GPB. Evidence shows that high-quality EK may not always result in corresponding environmental actions. This is also relevant to the integrity of online media coverage of environmental issues [94]. Given the above, the following hypotheses were developed:

H5a. 

EK moderates the relationship between MP and EA.

H5b. 

EK moderates the relationship between PIM and EA.

A diagram illustrating the framework model used in this study can be seen in Figure 1.

3. Research Methodology

3.1. Instrument Development and Measures

The questionnaire was developed in strict accordance with ethical standards. Respondents read detailed information about the research study before they completed the questionnaire. The survey was anonymous and voluntary. Data were obtained only to assess the characteristics of the population at a given time. The questionnaire was prepared following Chinese legislation: the 20 August 2021 Personal Information Protection Law of the People’s Republic of China.

This study sought to analyze the influence of external information transmission on the environmentally conscious shopping habits of Chinese consumers. This study considered five variables, and the measurement questions for each variable were derived from established scales in existing research, with appropriate modifications made to suit context-specific consumer purchase scenarios. The MP scale was based on the Media Exposure to Environmental Information scale in [62], which has a unidimensional structure consisting of 4 measurement items. PIM was derived from a unidimensional scale with 4 measurement items [99]. The EA scale was based on [91] with 6 measurement items. The EK scale was based on [100,101] with 6 items. The measurement of GPB was taken from a 6-item measurement questionnaire developed by [102]. The structure of the questionnaire and the items are indicated in Appendix A.

Zeynalova and Namazova (2022) [103] discovered that the 5-point Likert scale was stable and accurate enough to calculate the questionnaire items in their study. Consequently, a 5-point Likert scale was employed for all of the aforementioned variables. The scale spanned from ‘1’ representing ‘strongly disagree’ or ‘never’ to ‘5’ representing ‘strongly agree’ or ‘always’. The measurement tools mentioned above have been used in authoritative studies to ensure the scale’s content validity. To ensure consistency between Chinese and English, the ‘translation-back-translation’ technique was employed [12]. This process entailed the translation of the English questions into Chinese, followed by their subsequent translation back into English for the purpose of comparison. Four experts were invited to validate the English and Chinese questionnaires to guarantee the privacy of the respondents during the questionnaire process and promise that they could stop their participation at any time.

3.2. Critical Organisms of Selecting Eco-Friendly Clothing

To give respondents the ability to more critically consider their behaviors when filling out the questionnaire, the questionnaire was designed to select a green product that was an everyday necessity in their lives, thus better activating the SOR framework [22]. In this study, eco-friendly clothing was chosen as a critical green product. The reasons for this were as follows: Firstly, the promotion of eco-friendly clothing is a key part of achieving green sustainability [89]. Secondly, consumers may eventually give up buying eco-friendly clothing after complicated thinking during actual purchases [104]. Finally, the literature suggests that the market potential for eco-friendly garments in China is increasing and the research intensity is rising [1]. This is in line with current critical needs.

3.3. Sample and Data Collection Procedures

This study investigated the influence of MP and PIM among netizens on customers’ GPB in the context of China’s fast-paced new media growth. It did not impose any geographical restrictions due to the transactional and pan-geographical nature of e-commerce [105]. To avoid measurement bias caused by subjective assumptions [106] and circumvent perception differences caused by some groups with prejudice against online purchasing [107], this study only included Chinese consumers with internet purchasing experience.

The tests used to select valid surveys were as follows: Firstly, samples with social approval bias or subjective cognitive bias toward green consumption were excluded based on the logical relationship between the average score of green purchasing behavior and the screening question items. Secondly, respondents who were not an audience of new media communication were excluded. Thirdly, respondents who were younger than 18 years old were excluded. Finally, incomplete questionnaire answers were excluded.

The questionnaires were distributed through the professional data collection platform ‘Credamo (www.credamo.com)’ in China. This platform is used for academic and social surveys and has established strategic partnerships with several Chinese universities and enterprises [1]. The official distribution period for the questionnaires was from 8 January to 19 March 2024. In this study, simple random sampling was employed, resulting in the issuance of 450 online questionnaires during the survey period. Of these, 357 were deemed valid, yielding an effective response rate of 79.3%.

The demographic characteristics of the sample are presented in

Table 1. Statistical analysis of respondents’ characteristics.

Construct (N = 357)	Group	Frequency	Percentage (%)
Gender	Male	147	41.2%
	Female	210	58.8%
Age	18–24	103	28.85%
	25–34	142	39.78%
	35–44	68	19.05%
	Over 44	44	12.32%
Individual monthly income	3000 and below	68	19.05%
	3001–6000	54	15.13%
	6001–9000	66	18.49%
	9001–12,000	66	18.49%
	More than 12,000	103	28.85%
Educational level	High school and below	10	2.80%
	Diploma	20	5.60%
	Undergraduate	254	71.15%
	Master	73	20.45%

. Of the total number of participants, 147 (41.2%) were male and 210 (58.8%) were female. The age group comprising individuals between 25 and 34 years of age accounted for the largest proportion of participants, representing nearly 40% of the total. The majority of participants (over 70%) possessed a Bachelor’s degree. This indicates that individuals with higher levels of education tend to demonstrate a heightened awareness and concern regarding environmentally conscious consumption practices. The distribution of income was relatively balanced, with the highest proportion belonging to the high-income group (those with an income of more than RMB 12,000), which accounted for nearly 30% of the total.

3.4. Methods of Analysis

The normality of the questionnaire data and the common method bias were tested. The data were subjected to analysis in three distinct stages. The reliability and validity of the sample data were evaluated. The reliability and validity of the scales were evaluated through the utilization of Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) to ascertain their stability. This entailed an assessment of structural, convergent, and discriminant validity [108]. Subsequently, a latent variable SEM was constructed to conduct path tests to validate the direct and mediating effects of the model. Furthermore, the study employed the LMS method to assess the moderated and moderated mediated effects of the model [109].

Specifically, this paper presents the construction of structural models to test the significance of the effects of MP and PIM on EA, MP and PIM on GPB, and EA on GPB. Subsequently, the mediating effect of EA is examined. Finally, interaction terms are constructed to test the moderating effects of EK on the independent and mediating variables and the moderated mediating EA.

4. Data Analysis and Results

4.1. Normality Test

The normality test of the sample data could avoid bias in the SEM [58], and the questionnaire data were analyzed for normal distribution through SPSS27.0. The results are presented in

Table 2. Kurtosis and skewness.

Contact	Mean	Standard Deviation	Skewness	Standard Error	Kurtosis	Standard Error
MP	3.0413	0.73883	−0.086	0.129	−0.648	0.257
PIM	3.8487	0.61227	−1.068	0.129	1.509	0.257
EA	4.1569	0.59478	−1.397	0.129	2.161	0.257
EK	3.5840	0.72944	−0.548	0.129	−0.025	0.257
GPB	4.1508	0.53896	−1.05	0.129	1.277	0.257

.

Table 2 shows that the maximum absolute value of univariate skewness was 1.39, with some values of less than 1, and the maximum absolute value of univariate kurtosis was 2.1, with most values being less than 2. The sample data were tested and found to be free of outliers. According to [110], the value of skewness should not be exceed −2 to +2 and the value of kurtosis should not be over −7 to +7. The sample data’s multivariate normal distribution was confirmed and the parameter estimation using structural equation modeling was robust [111].

4.2. Common Methodological Biases

Since the data in this study were collected by a questionnaire survey that relied on the respondents’ accounts, it is possible that the connection between the variables could have been affected by the shared technique. To minimize any misunderstandings, this study carefully revised the question items of each variable based on a mature scale before conducting the questionnaire. Measures were taken to anonymize the measurement and set the order of the questions reasonably. However, procedural control could not eliminate common method bias. Hence, testing for any common technique bias in the measurement data was crucial.

The data of all latent variables were analyzed using factor analysis in SPSS 27.0. The analysis revealed four factors with eigenvalues greater than 1. The first factor explained 39.918% of the variance, which was less than 40% and conformed to Harman’s test. A one-way validated factor analysis was used for re-validation. The results were as follows: χ2/df = 4.154 > 3; RMSEA = 0.094 > 0.08; GFI = 0.713 < 0.9; RMR = 0.059 > 0.05; CFI = 0.785 < 0.9; IFI = 0.786 < 0.9; and TLI = 0.766 < 0.9. The model’s fitted parameters were inadequate, indicating the absence of any significant methodological bias in this study.

4.3. Reliability and Validity Tests

4.3.1. Reliability and KMO Test

The reliability of the scales was tested using Cronbach’s alpha value in SPSS27.0.

Table 3. Kaiser–Mayor–Olkin (KMO) and Bartlett test.

Construct	Cronbach’s Alpha (CA > 0.7)	KMO Value (KMO > 0.6)	Bartlett Test Significance (p < 0.05)	Factor Analysis Suitability
MP	0.754	0.737	<0.001	Acceptable
PIM	0.749	0.766	<0.001	Acceptable
EA	0.866	0.871	<0.001	Acceptable
EK	0.885	0.903	<0.001	Acceptable
GPB	0.829	0.863	<0.001	Acceptable

indicates that the alpha coefficients of each scale were above 0.7, indicating that the questionnaire reliability passed the test [108].

The sample data were analyzed using SPSS27.0 and subjected to KMO and Bartlett tests. Table 3 shows the presence of KMO values above 0.7 and Bartlett test p-values below 0.05, suggesting that the sample data were appropriate for conducting additional factor analysis [112].

4.3.2. Validity Tests

The items used in this study were taken from established scales in authoritative journals, ensuring good content validity [113]. CFA was then conducted to test the structural validity of the questionnaire [114]. The fit indicators of the model are in

Table 4. Main indices of model fit test.

Fitness	${\mathit{X}}^2/\mathit{d}\mathit{f}$	RMSEA	CFI	IFI	TLI	SRMR
Reference	1 < NC < 3	$0.08$	$\ge 0.9$	$\ge 0.9$	$\ge 0.9$	$0.1$
Value	1.872	0.049	0.942	0.943	0.935	0.052

. The modeling fit was 1 <$X^2/df$ = 1.872 < 3, RMSEA = 0.049 < 0.05, and SRMR = 0.0519 < 0.1. All of the remaining indicators were above 0.9, indicating a strong fit of the model and excellent structural validity of its scales.

The following section explores the convergent validity of the sample data. Convergent validity refers to the extent to which various measurements of a certain concept exhibit a strong correlation [115]. This study validated this by measuring several key indicators. Firstly, the standardized factor loading (SFL) needed to meet the criterion that the factor loadings of the items in the model should be equal to or larger than 0.5 [115,116]. Secondly, the composite reliability (CR) of the model [117] indicates that a CR above 0.6 is an acceptable threshold. The third consideration was the average variance extraction (AVE) of the proposed model, which had a minimum threshold of 0.5 [116,117].

Table 5. Convergent validity results.

Constructs	Items	SFL	TVE	CR	AVE > 0.5
MP	MP1	0.809	58.039%	0.846	0.580
	MP2	0.835
	MP3	0.695
	MP4	0.698
PIM	PIM1	0.764	57.86%	0.846	0.579
	PIM2	0.803
	PIM3	0.741
	PIM4	0.732
EA	EA1	0.764	60.117%	0.900	0.601
	EA2	0.751
	EA3	0.793
	EA4	0.82
	EA5	0.82
	EA6	0.697
EK	EK1	0.748	63.780%	0.913	0.638
	EK2	0.793
	EK3	0.824
	EK4	0.787
	EK5	0.798
	EK6	0.838
GPB	GPB1	0.742	54.341%	0.876	0.544
	GPB2	0.809
	GPB3	0.758
	GPB4	0.719
	GPB5	0.801
	GPB6	0.569

demonstrates the results.

Table 5 below shows that the minimum value of the SFL for all latent variables was 0.569 > 0.50, the lowest CR was 0.846 > 0.7, and the lowest AVE was 0.544 > 0.5, all of which were much larger than the critical values of the indicators, providing evidence that the scale exhibited strong convergent validity.

In addition, discriminant validity refers to the degree to which variables differ. The arithmetic square root of the AVE value for the construct was greater than the Pearson value with the corresponding variable [116], and the Pearson coefficient was less than 0.9, indicating good discriminant validity [118].

Table 6. Assessing discriminant validity.

	MP	PIM	EA	EK	GPB
MP	0.762
PIM	0.423 **	0.761
EA	0.455 **	0.727 **	0.775
EK	0.541 **	0.604 **	0.564 **	0.799
GPB	0.437 **	0.644 **	0.732 **	0.581 **	0.738

Note: Bold diagonal values are square roots of AVE; lower triangles are Pearson correlation coefficients between variables. ** p < 0.01.

confirms that the discriminant validity between the constructs passed the test.

4.4. SEM Results

This paper presents an SEM based on the theoretical framework. The overall fit index of the model was $X^2/df$ = 2.162, RMSEA = 0.057, GFI = 0.907, CFI = 0.936, IFI = 0.937, and TFI = 0.926. The model fit was good. The results of the SEM in this study are shown in Figure 2.

Figure 2 and

Table 7. Hypothesis testing (direct effect).

H	Path	Estimate	S.E.	C.R	p
H1a	MP → EA	0.219	0.044	4.365	***	Supported
H1b	MP → GPB	0.144	0.043	2.507	0.012	Supported
H2a	PIM → EA	0.852	0.081	9.246	***	Supported
H2b	PIM → GPB	0.293	0.105	2.051	0.040	Supported
H3	EA → GPB	0.553	0.125	3.739	***	Supported

*** p < 0.000.

show the results of hypothesis testing. The path MP → EA (β = 0.22 ***) indicated that MP significantly influences EA. H1a was confirmed. Path MP → GPB (β = 0.14 *) showed that MP significantly influences GPB. H1b was supported. Path PIM → EA (β = 0.85 ***) indicated that PIM significantly influences EA. H2a was supported. Path PIM → GPB (β = 0.29 *) showed that PIM significantly influences GPB. H2b passed the hypothesis test. Path EA → GPB (β = 0.55 ***) showed that EA significantly influences GPB. H3 passed the hypothesis test.

Furthermore, the findings of the hypothesis test showed that H1b and H2b had a high level of statistical significance. Low p-values suggested that online media publicity does not strongly influence GPB. This finding aligns with the fundamental principle of the SOR, which posits that external stimuli can shape individuals’ ideologies but do not necessarily translate into actual purchasing behaviors [22]. To study the impact of the MP and PIM on GPB using the SOR, it was important to analyze the mediating effects of the model.

4.5. Mediating and Moderating Effects

4.5.1. Mediating Effects of Environmental Attitude (EA)

This study analyzed the mediating role of EA. Based on the SEM, the mediating effects were analyzed using the bias-corrected non-parametric percentile bootstrap method, and the number of samples was set at 5000 [119]. The results are indicated in

Table 8. Test for mediation effect.

Hypothesis	Path	Estimate	S.E.	Bootstrap 95%CI		p
				Lower	Upper
H4a	MP → EA → GPB	0.094	0.046	0.024	0.204	**
H4b	PIM → EA → GPB	0.363	0.161	0.113	0.747	**

** p < 0.01.

.

Table 8 indicates that the mediation path (MP → EA → GPB) of value was 0.094 (**) and the confidence interval [0.024, 0.204] did not contain 0, demonstrating that the partial mediation effect was significant. This showed a partial mediation effect of EA between MP and GPB. H4a was verified. The mediation path (PIM → EA → GPB) of value was 0.363 (**) and the confidence interval [0.113, 0.747] did not contain 0, so the partial mediation effect was significant. This showed a partial mediating effect of EA between PIM and GPB. H4b was supported.

4.5.2. Moderating Effects of Environmental Knowledge (EK)

The moderating effects analyses of the models were conducted using Mplus 7.0 for SEM, and the research procedures adopted the LMS method provided by [108]. Firstly, for the SEM (model 1) without latent moderation (interaction) terms, the model fits were χ²/df = 1.978, RMSEA = 0.052, CFI = 0.926, and TLI = 0.917. Model 1 was well fitted (1 < χ²/df < 3, RMSEA < 0.05, and the rest of the fitness indices were all greater than 0.9), with AIC = 21887.651 and H0 = −10810.923.

Then, the interaction term-moderated model (model 2) was built and the goodness-of-fit parameters of the moderated model were obtained as AIC = 18823.777 and H0 = −9319.888. The calculation showed that the AIC of model 2 was smaller than that of model 1 by 2996.069, and the chi-square test for the value of −2LL was significant (***). Thus, the moderated model (model 2) had better results than model 1, with good goodness of fit [120].

The results found that the interaction term of MP and EK (MP XWITH EK) on EA passed the significance test (β = −0.089 **), suggesting that EK has a negatively moderating effect between MP and EA. H5a was supported. The interaction term between PIM and EK (PIM XWITH EK) significantly affected EA (β = −0.208 ***), indicating that EK negatively moderates the relationship between PIM and EA. H5b was confirmed.

To enhance the comprehensibility and logical structure of the text, this study employed the point selection method to classify EK into three groups: low (M − 1SD), medium (M), and high (M + 1SD). The effects of MP and PIM on EA were analyzed under high or low levels of EK. Based on this analysis, a simple slope analysis diagram is presented.

In Figure 3, the steep linear slope indicates that MP strongly affects EA when EK is low. However, this effect diminishes under high EK, with MP having a weaker positive impact on EA. Similarly, Figure 4 demonstrates that PIM has a relatively weak positive influence on EA when EK is high, as evident from the flatter slope. Conversely, when EK is low, PIM exerts a more significant positive influence on EA, as shown by the steeper slope.

4.5.3. Moderated Mediation Effects

Moderated mediation occurs when the mediating effect is influenced by the moderating variable. This can be verified by constructing the mediating effect in the full model [121]. The goodness of fit of model 2 has been previously demonstrated by constructing the parameters of interest and using maximum likelihood estimation with a confidence level of 95%. The results were calculated using MPLUS 7.0, as shown in the table below.

From

Table 9. Moderated mediation effects (MP → EA → GPB).

Mediator	EK	Mean	SE	LLCI	ULCI
EA (MP→EA→GPB)	High	0.080 (**)	0.028	0.026	0.135
	Medium	0.107 (**)	0.037	0.035	0.179
	Low	0.133 (**)	0.047	0.040	0.226
Contrast	High–Low	−0.053 (*)	0.025	−0.013	−0.003

** p < 0.01, * p < 0.05.

, irrespective of the subgroup within which EK was situated, the estimates for path MP→EA→GPB were statistically significant, and none of the confidence intervals encompassed the value of 0, indicating the robustness of EA’s mediating role. Additionally, pairwise comparisons of the mediating effects across the different levels of EK revealed that the differences in the mediating effects were significant, as their confidence intervals did not contain 0. This demonstrated the presence of a moderated mediating effect. Specifically, as the level of EK increased, the mediating effect of EA between MP and GPB weakened significantly. This indicates that EK negatively moderates the relationship, reducing the strength of EA’s mediating effect as EK rises.

Similarly, according to

Table 10. Moderated mediation effects (PIM → EA → GPB).

Mediator	EK	Mean	SE	LLCI	ULCI
EA (PIM→EA→GPB)	High	0.284 (**)	0.083	0.121	0.447
	Medium	−0.062 (*)	0.026	−0.114	−0.010
	Low	0.407 (**)	0.129	0.155	0.660
Contrast	High–Low	−0.124 (*)	0.053	−0.227	−0.020

** p < 0.01, * p < 0.05.

, the mediating effect of EA with moderation between PIM and GPB still held. The mediating effect of EA showed a significant weakening trend as the level of EK rose. That is, the mediating role of EA between PIM and GPB was negatively moderated by EK.

5. Discussion

The results of this study support H1a, H1b, H2a, H2b, and H3, indicating that PIM and MP have a stimulating effect on consumer EAs and GPBs during the marketing process. This is similar to a prior study by [2,12], which demonstrated that the publicity of external environmental information can effectively influence consumers’ purchasing decisions. MP and PIM as drivers of EA enable consumers to re-conceptualize and re-evaluate the value of green products [66,75]. As individuals become more environmentally aware, whether influenced by official media or word-of-mouth information, consumers participate in GPB to achieve sustainable social development [17].

Furthermore, a comparison of the data results for MP and PIM reveals that consumers are more inclined to accept the spontaneous word-of-mouth publicity disseminated by internet users than that disseminated by traditional media. The declining credibility of the official media is further substantiated [80].

H4a and H4b were both supported, suggesting that the relationship between MP, PIM, and GPB is mediated by EA, consistent with the findings of [91,92]. This study suggests that consumers are initially influenced by external environmental information to develop more positive EAs. Similar to [87], the influence of such informational stimuli does not necessarily translate directly into GPB.

Whilst both MP and PIM contributed to EA, PIM had a greater impact on GPB through the mediating role of EA. This underlines the key role of IWOM marketing in driving sustainable consumer behavior. These findings are similar to [12]. The significant mediating role of EA in the relationship between PIM and GPB suggests that consumers place greater trust and credibility on information obtained from peers and social networks than from official media sources. This may be due to the decline in the credibility of the media [80] or the development of a new media network environment [74]. By doing so, they can effectively nurture EA and encourage sustainable purchasing decisions [84].

Finally, the establishment of H5a and H5b demonstrates the negative regulatory role of EK. This again validates the results of [122], while Hamzah [98] found that EK positively moderates the relationship between perceived green value and green purchase intentions. This suggests that EK enhances an individual’s subjective judgment and that an individual’s attitude does not change easily when stimulated by environmental information [123]. In addition, EK has a negative moderating effect on the mediating effect of EA. The higher the level of EK, the weaker the predictive validity of EA on GPB, thus reducing the mediating effect. This is in line with [2,124], who found that the phenomenon of desensitization, whereby consumers show counteracting effects when exposed to frequent environmental information stimuli, weakens the transformation of EA.

6. Conclusions

This study aimed to uncover the formation mechanisms of GPB of Chinese e-commerce consumers in the face of the influence of environmental information from different channels. The respondents were all e-commerce consumers from China. The SOR and persuasion theories were combined to predict the direct relationship between MP, PIM, EA, and GPB, and the mediating effect of EA as well as the moderating utility of EK were examined. The results indicated that both MP and PIM positively influence EA and GPB in the new media environment, supporting hypotheses H1a, H1b, H2a, H2b, and H3. EA plays a partial mediating role in this process, confirming H4a and H4b. Additionally, all of these utilities are moderated by EK, with H5a and H5b being supported. The findings confirm that media campaigns and IWOM are indeed important in developing positive EAs and further transforming them into green behaviors. This is of great theoretical and practical significance for encouraging and guiding GPBs and promoting environmentally sustainable development.

6.1. Theoretical Contributions

This study employed a combination of persuasion theory [41] and SOR [26] to investigate the driving mechanism of GPB. This approach served to complement the existing theoretical system. Additionally, this study incorporated the influence of new media marketing in China, with a focus on media information [12] and IWOM [12]. This distinction allowed for a more nuanced understanding of the factors influencing GPB compared to previous studies that primarily focused on the influence of a single advertisement appeal on green behavior. The combination of external environmental information demand and the SOR clarifies the transformation process of consumers after being stimulated by external environmental information, which, in turn, promotes GPB [39]. Furthermore, this study provides theoretical support for the government in formulating green policies and for companies in marketing. It contributes to the achievement of sustainable development strategies.

6.2. The Potential Implications for Managerial Practice

This study examined the response and transformation mechanism of consumers in receiving environmental information propaganda. It included data analysis and empirical tests, which are of practical significance to the government and enterprises in motivating and guiding consumers to practice green behaviors.

6.2.1. From a Governmental Perspective

(1) The government can issue pertinent legislation and regulations to enhance the monitoring system and prevent certain media outlets from being unduly influenced by commercial interests to disseminate misinformation about environmental matters, which could ultimately prove detrimental to consumers. Nevertheless, while it is crucial to guarantee the veracity and dependability of media content, the viability and consequence of government monitoring must also be meticulously evaluated. Enhanced transparency and accountability within media organizations, in conjunction with the formation of autonomous regulatory bodies, may prove more efficacious than direct monitoring. Alternatively, when media entities are autonomous and disassociate their affiliations with government or business, their credibility with the public is likely to increase.

(2) It is recommended that the e-commerce law be further refined and effective management policies be introduced in light of the latest developments in e-commerce platforms. It is therefore necessary to create a favorable shopping environment and provide consumers with a superior consumer experience. It is important to prevent merchants from exploiting the public’s tendency to assume responsibility to create a false positive IWOM.

(3) The preceding conclusions demonstrate that consumers are positively influenced by media information and internet word-of-mouth concerning their environmental attitudes and behavior. It is recommended that the government utilize the internet to encourage green purchasing behavior [16]. This could be achieved by increasing the frequency of publicity, education, and subsidies to encourage green moral exhortation and value dissemination. Nevertheless, it is crucial to exercise caution when interpreting the results, as the methodology employed may have introduced some degree of bias. Furthermore, it would be beneficial to have a more precise understanding of the audience groups in question.

6.2.2. From a Corporate Perspective

(1) Enterprises should initially prioritize the greening of their products. It is recommended that a process for the green practice of the product throughout its entire life cycle be established. The focus should be on the quality of green products and services, as this will foster brand identity and enhance user satisfaction. As consumer environmental awareness increases, they will consider the environmental attributes of products when purchasing, provided that these attributes do not compromise their own consumption needs [125]. It is not possible to sacrifice product features, costs, and other aspects in exchange for the green attributes of the product.

(2) Enterprises rely on the internet to establish a comprehensive marketing system. On the one hand, the utilization of green publicity reinforces the brand image and environmental identity, coupled with the introduction of practical green services and green marketing tools. Conversely, the quality management of green marketing must be enhanced, and the appointment of specialists to monitor media information and internet word-of-mouth in real-time is essential to prevent the phenomenon of “greenwashing”, which could have a profoundly negative impact on enterprises.

(3) The establishment of technical standards and environmental certification systems, in conjunction with the expansion of sales and publicity channels, serves to ensure that consumers can easily identify and purchase green products and services that meet their needs, thereby promoting the active practice of green purchasing behavior.

(4) When disseminating environmental information via the media and the internet, it is essential to have accurate user profiles, block the various mechanisms that result in a loss of interest in green purchasing behavior, and utilize the transformation mechanism of environmental attitudes to its fullest extent. For those with a high level of environmental knowledge, it is important to not only publicize the impact of environmental degradation but also clearly state the high likelihood of such harmful effects. This is to prevent consumers from questioning and resisting.

7. Research Limitations and Future Research Directions

It should be noted that this study is not without limitations. Firstly, it should be noted that the data were obtained through respondents’ self-reports, which may have introduced common methodological logical biases. Secondly, it is unfeasible to entirely circumvent the detrimental impact of social trends on the outcomes of data analysis. China is a society characterized by a culture of face-saving, group pressure, and a tabula rasa approach to governance, which can result in respondents presenting their actual thoughts in a manner that is socially acceptable. Thirdly, the research hypotheses were based on users’ ideal information environment, but people’s actual shopping environment may be affected by unknown challenges, such as the simultaneous stimulation of positive and negative information. Fourthly, the operation strategies of different shopping platforms may also affect consumers’ decisions and create information bias. Finally, the impact of demographic variables was not taken into account. For instance, individuals of different ages may have a different level of awareness of information networks, while those with higher education may be more inclined to pay attention to environmental information. Additionally, geographic location differences may influence the perceived value of information among respondents who are exposed to minimal environmental pollution.

In future research, the interplay of positive and negative environmental information should be considered in the actual shopping environment [22]. Information bias due to the dynamic pricing mechanism of different platforms should be further explained in the purchase decision [125]. Additionally, future studies should give greater consideration to the information bias caused by demographic factors to improve the findings.