Introduction

According to a study conducted by the Global Carbon Project published in the journal Earth System Science Data, global carbon emissions will exceed 40.9 billion tons in 2023. In total, carbon emissions from fossil fuels are expected to rise again, reaching 36.8 billion tons, an increase of 1.1% compared with 2022 emissions (Liu et al., 2024). In recent years, governments around the world have become increasingly aware of the problems caused by carbon emissions. Major developed countries and organizations such as the United States, Canada, South Korea, Japan, and the European Union have pledged to achieve carbon neutrality by 2050. In 2020, the Chinese government pledged to achieve a carbon peak before 2030 and carbon neutrality before 2060, formulating action plans to optimize the industrial structure and energy structure in 2021 (Li et al., 2024). In the context of the constraint of the “dual carbon” target and the promotion of the development of clean rural energy industry, it is particularly important for China to increase the use of clean rural energy (Song et al., 2023; Yin et al., 2024).

One vulnerable group in society is those who live in China’s rural areas generally who have low incomes. Therefore, it may be more difficult to achieve a clean energy transition in rural areas than in urban areas. There is a significant gap in the energy structures of urban and rural households in China. The traditional energy sources often used by farmers include biomass (firewood, straw), petroleum, coal, etc. Coal has long been the main energy source for farmers in China (Liu et al., 2023). However, the use of traditional energy sources leads to environmental pollution, excessive carbon emissions, and other problems. Consequently, the state has issued policies, especially those focused on rural areas, for the development of clean energy and the vigorous promotion of clean energy use. The 14th Five-Year Plan for Promoting Agricultural and Rural Modernization was issued to promote the development of photovoltaic power, wind power, and other new energy sources in rural areas (Liu et al., 2023; Yu and Yin, 2023). New models and new forms of renewable energy have been extensively developed. With the strong promotion of national policies, clean energy projects have been established in some northern regions, and a certain amount of progress has been made at the macro level. However, there are still significant differences in the policy direction, degree of attention, and development objectives of China’s rural clean energy plans, and farmers’ enthusiasm for clean energy is not high. To transform and upgrade the rural energy structure and realize the efficient use of clean energy, it is necessary to rely on the “cluster effect” generated by hundreds of millions of farmers’ clean energy use in production and everyday life (Yu and Yin, 2023).

Farmers are the main body responsible for rural energy use, as well as the most direct actors, the main body of decision-making, and the ultimate beneficiaries. Many scholars have focused on the impact of government actions, as well as farmers’ income levels, cognition, and livelihood modes, on their willingness to use clean energy (WTUCE). Answering the question of how to effectively guide farmers to adopt clean energy has become the key for rural families realizing clean energy use. Compared with traditional energy, the value of clean energy is reflected mainly in the interaction between people and the natural environment. In fact, the adoption of clean energy by farmers is a systematic decision-making process, including premise cognition, intermediate decision-making, concrete adoption, and other links, whereas the cognition of clean energy use is at the root of the entire decision-making process (Zhang et al., 2023; Yin et al., 2024). Therefore, whether rural families adopt clean energy largely depends on the ecological cognition (EC) level of the farmers (Li and Wang, 2023). In this study, EC was selected as a variable affecting farmers’ WTUCE. The key factor at work in improving farmers’ EC level is the promotion of their adoption of clean energy by improving their emotional resonance and strengthening their ecological moral self-discipline. Transforming farmers’ EC into production behaviors with environmental awareness is a strategy that is fundamental to China’s agricultural green transformation (Li et al., 2024). According to the hypothesis of the “rational economic man”, farmers, as rational people, decide whether to adopt clean energy or not according to their own cognition of its ecological value (Liu et al., 2019). However, farmers’ perceptions of ecological value depend on information about EC to which they are exposed. With the rapid development of internet technology, new media such as TikTok, Weibo, WeChat, and Kuaishou have become increasingly important means by which people can obtain information. Farmers can obtain information conveniently and quickly to change their behavior on the basis of new media use (NMU). New media publicizes the current events taking place in various countries and regions around the world every day. Farmers receive all kinds of information every day. Some false information confuses farmers’ cognition and leads them to hold a neutral attitude toward media information. Therefore, media trust (MT) affects farmers’ EC obtained from new media, which further affects new farmers’ WTUCE.

The existing studies in this area have focused mainly on national policies, rural infrastructure construction, farmers’ income, farmers’ education levels, and village topography as variables affecting WTUCE (Zhang et al., 2023; Liu et al., 2019; Elahi et al., 2022). Studies have shown that the stronger the national policy is, the better the infrastructure; the higher the farmers’ income and education level are, the stronger the WTUCE (Elahi et al., 2022). Some scholars have also studied policy perception intensity and demographic characteristics as variables affecting WTUCE, and these two variables significantly and positively affect the WTUCE. Some scholars have posited that EC affects the WTUCE, and the relevant research results show that each dimension of EC has a significant positive effect on clean energy use. Moreover, the impact of EC on WTUCE is heterogeneous across different landforms (Xie et al., 2021). A number of studies have explored the factors influencing farmers’ WTUCE, but few studies have investigated the impact of NMU on farmers’ WTUCE. Although existing studies have taken EC as a variable affecting farmers’ WTUCE, they have not considered MT as a moderating variable to explore the impact of NMU on farmers’ EC, nor have they proven that NMU affects the WTUCE through the intermediary variable of EC (Lin and Li, 2022).

Therefore, this study aimed to build a theoretical model of farmers’ WTUCE from a micro perspective, exploring the impact of NMU on farmers’ WTUCE, as well as the mechanism of the two specific variables of MT and EC. Moreover, we also conducted a heterogeneity analysis to explore whether different terrains and different nonagricultural income ratios had different effects on farmers’ WTUCE. We aimed to achieve the following research objectives. (a) To quantify NMU by rural residents and explore the influence mechanism of NMU on farmers’ WTUCE. (b) To explore the mediating role of EC and the moderating role of MT. (c) To determine whether NMU by farmers promotes or inhibits their WTUCE. (d) To explore the differences in the test results across different terrains and different levels of nonagricultural income. In addition, this study makes the following research contributions: (a) It explores the internal mechanism by which NMU impacts farmers’ WTUCE. (b) Different suggestions are provided according to different terrain types and different proportions of nonagricultural income. (c) It assists in the construction of an ecological civilization in China and promotes the development of rural revitalization.

The remainder of this paper is structured as follows. Section “Theoretical basis and research hypothesis” reviews the literature on the four variables and proposes research hypotheses. Section “Methodology” describes the research design. In the section “Analysis of empirical results”, we present the empirical research and analysis of the results. In the section “Further analysis and discussion”, the two variables of terrain and nonagricultural income share are analyzed and discussed. Section “Conclusions and future prospects” presents our conclusions.

Theoretical basis and research hypothesis

Theoretical basis

The WTUCE of farmers

The WTUCE of farmers mainly includes their willingness to use new energy vehicles, the clean utilization of straw, and their willingness to use low-carbon energy sources such as solar energy, heat energy, and wind energy (Lin and Li, 2022; Gui and MacGill, 2018; Lowitzsch et al., 2020). The main research on this topic is as follows. Xu (2021) reported that clean energy policies in rural areas have a significant effect on clean energy consumption. Dominguez et al. (2021) proposed the energy ladder theory, which suggests that, as incomes increase, farmers increasingly use cleaner fuels. Yu et al. (2022) suggested that more government investment and subsidies in new energy infrastructure would increase farmers’ WTUCE. Dai (2022) noted that adjusting the new energy price subsidy policy and improving tax measures would effectively promote the development of rural photovoltaic power, wind power, and other new energy sources. Min and Mayfield (2023) explained the different factors affecting the WTUCE, such as heat pumps, solar energy, and electric energy, but noted that the common factors are income and education levels. Chang et al. (2023) explained that, for different types of clean energy, the clean energy consumption of farmers depends, to a certain extent, on the family’s livelihood and the attractiveness of the village. Xie et al. (2023) proposed that the key to stimulating farmers’ adoption of clean energy is to guide their participation and stimulate their WTUCE. Li and Song (2023) proposed that the impact of increasing income on farmers’ use of clean energy varies by region and by the use of clean energy. The development of clean energy in rural areas can be achieved through the combined efforts of the government, enterprises, and villagers. Bashir et al. (2024) reported that resource availability, education level, and power failure are the key determinants affecting residents’ propensity to invest in renewable energy. The main factors influencing farmers’ WTUCE include farmers’ income, national policies, village topography, transportation conditions, policy cognition, the channels through which information is received, and farmers’ cognition level, all of which affect farmers’ consumption and use of clean energy.

New media use (NMU)

In recent years, the scale of China’s internet has expanded. With respect to NMU, different scholars have different definitions of new media. Studies have been conducted on the factors influencing NMU. There are also relevant studies on the relationship between NMU and the WTUCE.

The 52nd Statistical Report on the Development of the Internet in China shows that, as of June 2022, the number of Internet users in China was 1.051 billion, and the Internet penetration rate was 74.4%. Many kinds of internet applications continue to grow, and the user scales of instant messaging, network videos, and short videos have been ranked in the top three (Chi, 2023). Social media has also become an enormous source of various types of global and local news for millions of users (Aïmeur et al., 2023). Kuang (2012) noted that the core of new media is digital technology and interactive communication. Peng (2016) argued that new media refers mainly to interactive and integrated media forms and platforms based on digital and network technologies and other modern information technologies or communication technologies. Ma and Gao (2019) assessed the motivation factors of new media users, including four types of motivation: intellectual, emotional sustenance, entertainment, and expression. Tao and Zhu (2020) proposed that NMU is positively influenced by users’ social needs, cognitive needs, and self-presentation. Johnson et al. (2022) summarized the definition of social media as follows: social media is an internet-based channel that allows users to interact opportunistically and selectively to show themselves to audiences, and all gain value from user-generated content. Hua and Jin (2022) proved via an empirical analysis that there is a significant correlation between NMU and the use of traditional media, whereas the degree of users’ intellectual motivation has no correlation with their NMU. There have been few studies on the relationship between NMU and WTUCE. Zobeidi et al. (2022) constructed a structural equation model and concluded that social media affects WTUCE by influencing people’s perceptions. Li et al. (2023) proposed that new media can stimulate and improve enterprises’ green technology innovation. However, when farmers have the WTUCE, there is a deviation between their behavior and their willingness (Li et al., 2023). This study was informed by the concept of new media proposed by Ausat (2023), which includes interactive and social forms of media. In addition to new forms of social media, this category also includes MT aggregate short video platforms such as Douyin, Kuaishou, XiaoHongshu, and Bilibili. Roche et al. (2023) noted that the actual activities of social media have little effect on green energy companies. With respect to the distribution of questionnaires and the prediction of social development trends, we hypothesize that NMU positively affects WTUCE and verify this suggestion via empirical analysis.

Media trust (MT)

The literature shows that MT has a regulating effect on NMU, and people’s trust in new media is greater than their trust in traditional media. Moreover, MT can be classified from different perspectives. Turcotte et al. (2015) suggested that people’s trust in traditional media has declined, whereas social media has improved MT. As mentioned above, social media is one of the categories of new media, indicating that trust in new media has improved. Schranz et al. (2018) determined through a regression analysis that news consumption promotes trust in the media system. Strömbäck et al. (2020) concluded that trust in traditional media is declining and it is under attack from nonmainstream media. On the basis of factor analysis, Yang and Huang (2021) noted that urban residents’ trust in traditional media is higher than their trust in new media. From the perspective of the public, Knudsen et al. (2022) defined four aspects of MT: authenticity, thoroughness and professionalism, bias, and independence and objectivity. Riley and Robertson (2022) reported that the greater the proficiency of farmers in social media and the longer the period in which they have used it, the better their experience and the greater their degree of trust in the media. Jing and Liu (2023) divided MT into official MT and social MT, which correspond to different political attributes and value orientations. Pelau et al. (2023) investigated MT in terms of three main aspects: source credibility, information credibility, and media-type credibility. Shahbazi and Bunker (2024) argued that MT is highly important in network interactions, especially considering concerns about privacy and security. In summary, MT has a regulating effect on NMU. Users’ trust in media increases with their proficiency in using new media. MT includes three aspects: information credibility, the media type, and the credibility of the information source.

Ecological cognition (EC)

According to the literature, EC positively affects the green production activities of farmers and is conducive to green production. Bender (2020) emphasized that human cognition is a constraint that adapts to the task performance context. Chen et al. (2021) evaluated farmers’ understanding of eco-agriculture according to three different criteria: their understanding of eco-agriculture in terms of increasing income, saving water, and the price of ecological products. Sang et al. (2021) reported that EC has an intermediary effect on farmers’ green production activities under the influence of policy incentives, and it is conducive to farmers’ green production. Ren et al. (2022) reported that EC affects farmers’ understanding of the ecological environment and is a direct motivator of their behavior. Guo et al. (2022) suggested that farmers’ EC depends mainly on their understanding, perceptions, and action plans. Xu et al. (2023) reported that EC comprises farmers’ knowledge reserve of rural ecology, which is formed through their perception of the interdependence between agriculture and ecology. Tian et al. (2023) defined EC as farmers’ understanding of the ecological environment and ecological protection, as well as their understanding of the current status of environmental issues. Chen et al. (2023) suggested that farmers judge whether the ecological environment has improved after straw is put into the field based on their understanding of the ecological environment, ecological science, and green production. Zheng and Su (2023) noted that EC plays a positive role in ecological agricultural behaviors, such as mulch film recycling (Hou et al., 2019), organic fertilizer application (Zhang et al., 2020), and low-carbon technology (Li and Wang, 2023). In this study, EC was used to refer to farmers’ understanding of the current status of the ecological environment, environmental pollution, protection measures, and other aspects. Therefore, this study linked the variable of EC to WTUCE for hypothesis and verification.

In summary, many scholars have analyzed the factors influencing the WTUCE of rural families from the perspectives of income, education level, the ___location of a village, farmers’ cognition, government policies, etc. All the above factors affect the WTUCE of rural families. In view of this, this study explored the relationships between the three new variables, NMU, MT, and EC, and the WTUCE of farmers. Ultimately, it provides theoretical support and practical experience for the effective promotion of clean energy and energy transformation and upgrading.

Research hypotheses

New media use and willingness to use clean energy

Farmers’ attitudes toward using clean energy depend on their level of knowledge (Xie et al., 2023), which is influenced by the amount of information they have mastered. New media forms such as short videos and network broadcasts have the advantages of high levels of originality, strong interactivity, and a low threshold, which increase the duration and frequency of farmers’ use. With large amounts of information, rapid dissemination, low dissemination costs, and the characteristics of immediacy and interaction (Wang et al., 2023), new media enables farmers to obtain diversified information without leaving their homes and to participate in social discussions independently in ways that have a profound impact on their ideas. Therefore, new media can serve as an important link through which users can obtain information and improve their cognitive level. NMU can influence farmers’ WTUCE in several ways. One is the “information cocoon” effect (Li and Wang, 2023). New media platforms such as TikTok and Kuaishou can push more information about a subject according to people’s interests. When people click on videos related to clean energy, they show their interest in this field. The media platform will then continue to push information relevant to this topic. Farmers are exposed to increasingly concentrated information fields, forming an “information cocoon” about clean energy. This constantly strengthens farmers’ awareness of clean energy. Second, the interaction feature of new media has a positive effect on users’ behavioral willingness (Bushara et al., 2023). New media is a “two-way interactive” information exchange platform. Users can not only input knowledge in a one-way manner but also interact with new media to obtain a certain amount of recognition and feedback, which can increase their willingness to continue using it. When farmers use new media, they not only obtain information and cognition about clean energy from it but also publish their own knowledge of production and life, thereby enhancing their WTUCE. In addition, new media also has a spillover effect and a learning effect. Farmers who have used clean energy exert a demonstration effect on their neighbors, reducing the risk expectation and the learning cost of adopting clean energy among surrounding farmers. In light of the above analysis, Hypothesis H1 was proposed.

H1: NMU positively affects farmers’ WTUCE.

Ecological cognition and the willingness to use clean energy

According to the theory of planned behavior, the cognitive level of agents plays a decisive role in their behavior, and the will of the agents has a significant effect on their behavior (Feng and Zhao, 2023; Zhu et al., 2023). Individuals’ cognition determines their preferences and then influences their final behavioral decisions. Accurate EC is the basis of and a necessary prerequisite for farmers to form an understanding of green development and implement green production behaviors (Pelau et al., 2023), and it is also the basis for ecological protection behaviors. In this study, EC is understood mainly in terms of ecological protection policies, human impact perceptions of the environment, and whether participants know how to protect the ecological environment. According to the relevant literature, subjective norms and perceived behavioral control interact with each other and jointly determine the willingness of a person to participate. The influence of the cognitive level must be taken into account when analyzing factors related to farmers’ behaviors (Tian et al., 2023). Therefore, farmers’ EC level directly affects their willingness to protect the ecological environment, thus affecting their behaviors related to adopting clean energy. In terms of logic, this is represented as “EC → willingness → action participation”. According to the relevant theories, in the context of ongoing social development, the production and life of farmers are significantly affected by EC (Yao et al., 2016). The higher the level of farmers’ EC is, the more willing they are to protect the ecological environment (Qi et al., 2023). The WTUCE is also affected by EC. Some scholars have noted that EC plays a positive regulatory role in green agricultural production, but its contribution to the degree of green production is not high (Ren et al., 2022). In summary, when farmers have a clearer understanding of environmental risks and the ecological environment and are more aware of the ecological value of adopting clean energy, they are more inclined to use clean energy. Therefore, Hypothesis H2 was proposed.

H2: EC positively affects farmers’ WTUCE.

New media use and ecological cognition

New media has certain impacts on people’s lives, such as speeding up the pace of life, drawing people’s attention, and leading to timely thinking (Zhang, 2023). According to the literature, NMU has the characteristics of wide dissemination, fast dissemination, a large amount of information, and strong interactions. Moreover, the customer group of new media is very broad, meaning that users can obtain a variety of information and interact with more people. Cognition refers to the process of individual screening, organization, processing, and understanding of the obtained information (Chen et al., 2022), whereas EC refers to the process of individual screening and understanding of the obtained ecological information. In this study, EC refers to farmers’ understanding of the current status of the ecological environment, environmental pollution, and protection measures, among other aspects. With continued usage, new media pushes more information about a particular interest according to the user’s preference, forming an “information cocoon”. Therefore, when new media is used, it will repeatedly push information related to the ecological environment to farmers, and these farmers will screen and understand the acquired ecological information. This will cause them to think about ecological and environmental issues, and they may even take action, thus affecting their EC (Wang et al., 2023). The cognition of new things and new technologies can be influenced by social networks; when the new media has an impact on the EC of some farmers, the social network intensifies this impact, impacting the surrounding farmers and their relatives. Therefore, Hypothesis H3 was proposed.

H3: NMU positively affects farmers’ EC.

Media trust, new media use, and ecological cognition

According to the process of information transmission by new media, MT can be divided into information credibility, information source credibility, and media credibility. MT reflects the public’s dependence on the media and the information it releases, and the stickiness of users is also related to it. Generally, the more frequently a media platform is used, the higher the user’s trust in the media, and the higher the level of EC obtained from the media (Zafar et al., 2021). At present, NMU greatly exceeds the use of traditional media, and MT in traditional media is declining; meanwhile, trust in new media is increasing (Turcotte et al., 2015; Strömbäck et al., 2020), and trust in official media is greater than that in unofficial media. Currently, many official media outlets have opened accounts on platforms such as TikTok and Kuaishou; for instance, “News broadcast” and “People’s Daily” have accounts on Douyin. The release of information on new media platforms enhances users’ trust in new media and increases the stickiness of media users. When audiences use media information, they rarely separate by type of news sources, communication channels, or forms of news expression. The information is placed in an overall context as a comprehensive consideration. Moreover, the higher the users’ level of media proficiency is, the better their experience and the stronger their trust in the media (Riley and Robertson, 2022). Social networking sites are rooted in interpersonal relationships, and the greater their level of trust in the media is, the more users will continue to use these media (Zafar et al., 2020). Therefore, MT, as the audience’s comprehensive means of assessing news information, affects the audience’s perceptions, determining their interpretation of media information. Therefore, when farmers use new media, they judge their EC according to their general sense of the MT. The information presented by media outlets with high credibility can improve farmers’ EC, whereas the information produced by outlets with low credibility can worsen farmers’ EC. Therefore, Hypothesis H4 is proposed.

H4: MT has a moderating role in the relationship between NMU and EC.

Ecological cognition, new media use, and the willingness to use clean energy

On the basis of the literature review, EC is used to refer to farmers’ understanding of the current status of the ecological environment, environmental pollution, and protection measures, among other aspects (Pelau et al., 2023; Tian et al., 2023). New media, as a tool used to disseminate information to farmers, has the advantages of wide dissemination, high frequency, and fast speeds, and it has become an important way for farmers to obtain information in recent years. In the process of farmers using new media, new media pushes information related to clean energy to farmers. The effects of the “information cocoon” and “learning effect” will improve farmers’ EC and enhance their WTUCE (Li and Wang, 2023). The new media platforms display information related to clean energy in a more intuitive and vivid way, making it easier for farmers to understand, receive, and recognize it. Moreover, new media is a two-way interactive form of media. Farmers can communicate and provide feedback while receiving information so that farmers’ cognition of clean energy is shared among groups, and their EC is also improved in this interactive process. According to the theory of planned behavior (Ajzen, 1991), farmers’ volitional cognition determines most of their behavior. On this basis, EC plays a positive role (Zheng and Su, 2023) in determining ecological agricultural behaviors; thus, it can be concluded that EC affects the WTUCE. Finally, with the help of the information spread by new media, farmers’ ecological ideas and clean energy cognition spread to their neighbors and relatives via the “spillover effect”. It is the transmission mechanism of “NMU → EC → WTUCE”. Hypothesis H5 was proposed.

H5: EC has a mediating role in the relationship between NMU and WTUCE.

In summary, a theoretical model is shown in Fig. 1.

Fig. 1
figure 1

Theoretical model.

Full size image

Methodology

Questionnaire design

The questionnaire used in this study measures four main variables: NMU, MT, EC, and farmers’ WTUCE. To ensure the validity of the measured data, this paper carefully reviewed the relevant literature, integrated the acquired knowledge, and examined existing research findings when the measurement items for the scale were designed. A well-established scale suitable for this study was identified and appropriately modified to fit the research context. There are 4 potential variables and 20 items included in this measurement. Responses to the relevant questions were assessed via a five-point Likert scale. The measurement items are shown in Supplementary Information 1.

Study area

We used a detailed formal questionnaire survey to collect the data. These survey samples cover multiple provinces in northern China, with Hebei, Shandong, Henan, and Shanxi as the main concentration areas. In these four provinces, the pace of the construction of rural infrastructure and economic growth is relatively slow. The rural living conditions in these areas may be relatively harsh, and economic development is still in its early stages. Therefore, these areas’ development statuses better reflect the themes discussed in this article.

Survey and data collection

The method of identifying villagers was based on interpersonal networks, and questionnaires were distributed through social media and field visits. A questionnaire survey was used to collect the data, and the questionnaire content was designed on the basis of the research model. To ensure the reliability of the variables and questionnaire data, the authors conducted an extensive literature review, clarified the conceptual framework, and revised the existing scale on the basis of well-established scales in the literature and relevant knowledge. The scale was then tailored to align with the research direction of this study, quantifying respondents’ WTUCE. We also established a relationship model between farmers’ NMU, EC, MT, and WTUCE values.

The data used in this study were obtained via a network survey. The preliminary questionnaire was designed via the Questionnaire platform, and the network questionnaire was distributed to rural families on social media platforms such as WeChat and QQ via an interpersonal network. The questionnaire responses were collected online.

The questionnaire has four parts. The first part deals with the basic information of the respondents, including their gender, age, political status, annual income, and educational background. The second part addresses the independent variable of NMU and the moderating variable of MT, including the respondents’ NMU (usage time and years) and their trust in media (information, information source, and media). The third part addresses the intermediary variable of this study, i.e., EC, including the cognitive degree scale of 5 ecological and environmental items. The fourth part addresses the dependent variable of this study, i.e., the WTUCE of farmers, which is investigated mainly in terms of whether the respondents support the development of various clean energy sources. The SPSS 26.0 package was used for data analysis.

Analysis of empirical results

Descriptive statistical analysis

The survey period ran from January 30 to March 1, 2024. The questionnaire “Study on the Influencing Factors of New Media Use on farmers’ WTUCE” was released on the Juanxing platform and distributed through WeChat groups, WeChat moments, QQ, field visits, and other means, and a total of 300 questionnaires were collected. Excluding invalid questionnaires (unable to perceive environmental risk, incorrect content, or unusable), 263 valid questionnaires were recovered, and the sample recovery efficiency reached 87.7%.

Statistical analysis of the basic characteristics of samples

The basic personal information of the respondents includes their gender, age, education level, and annual income; the proportion of nonagricultural income in the farmers’ total income, whether they are Party members or village cadres; the topography of the region where they live; and whether they use clean energy. The specific results are shown in Table 1.

Table 1 Descriptive analysis of the demographic characteristics.
Full size table

As shown in Table 1, 114 respondents were male, accounting for 43.30%, whereas 149 were female, accounting for 56.70%. In terms of the age distribution, the respondents were mainly 18–30 years old and 41–50 years old, accounting for 34.2% and 34.8%, respectively, followed by 31–40 years old and over 50 years old, accounting for 11.4% and 12.9%, respectively, and there were only 7 respondents under 18 years old. The proportions of all the age groups in this sample are unbalanced. The reason for this may be that the visit was carried out during the winter vacation, and the young people had all returned to the countryside, so the number of samples in the 18–30 age group was large. In terms of education, 112 were college students or above, accounting for 42.6%, followed by 79 with junior high school education, accounting for 30%; high school or secondary school education and primary school education or below accounted for 18.6% and 8.7%, respectively. Each education segment is represented, which is in line with the age distribution characteristics of the sample. The 105 respondents with the largest proportion of non-agricultural income, accounted for 39.9%, and the proportions of the other stages were relatively similar. Among all the respondents, 84.4% were neither Party members nor village officials. The terrain of the areas where the respondents live is mainly comprised of plains, accounting for 72.6%, whereas other landforms accounted for ~9%. Among the respondents, 182 reported having used clean energy, whereas 81 believed that they had not. The proportion of those who have used clean energy is significantly greater.

Descriptive statistical analysis of model variables

A five-level Likert scale was used to measure the four potential variables, and the results are shown in Table 2. As shown in Table 2, the minimum value of each variable in the model is 1, and the maximum value is 5, but the mean values of each item are different, reflecting the differences in the respondents’ NMU, MT, EC, and WTUCE values. The lowest mean value was 3.51, which corresponds to the measurement item “understanding of rural ecological environmental protection policies” under the variable “EC.” This indicates that villagers believe that they do not have a deep understanding of rural ecological environmental protection policies. Another similar, smaller mean value was 3.57, which comes from the measurement item “Trust in the new media used” in “Media Trust”, indicating that villagers have low trust in new media. Although the rural clean energy information released by new media can be trusted, it may not be able to help villagers use clean energy. The maximum mean value was 4.19, which comes from “willing to use solar energy for photovoltaic power generation if conditions permit” in the variable “farmers’ WTUCE”, indicating that villagers have strong WTUCE. The absolute values of the skewness of the samples are all <3, and the absolute values of kurtosis are all <10, so the sample data basically conform to a normal distribution.

Table 2 Descriptive statistical analysis results of direction finding for each variable in the model.
Full size table

Reliability and validity tests

Reliability analysis

After testing, the Cronbach’s alpha value for the variable NMU was 0.873; for MT, it was 0.885; for EC, it was 0.885; and for WTUCE, it was 0.885. Reliability analysis, which is commonly represented by Cronbach’s alpha, was conducted to assess the consistency of the answers to the questionnaire. A Cronbach’s alpha value above 0.7 indicates good reliability and acceptable internal consistency. Since all the variables in this study had Cronbach’s alpha values above 0.7, they met the criteria for reliability and internal consistency.

Validity analysis

Content validity

Validity refers to the specific examination of the energy efficiency of each item and whether each item plays an important role. The validity test used exploratory factor analysis (EFA). When exploratory factor analysis results are sufficient, the Kaiser–Meyer–Olkin measure of sampling adequacy (KMO) is >0.6, and the Bartlett sphericity test is significant (P < 0.05). As shown in Table 3, the EFA results reveal that a total of four factors are extracted: NMU, MT, EC, and farmers’ WTUCE. After inspection, the KMO value was 0.929, and the P value was 0.000, indicating significance at the 5% level and indicating the reliability and accuracy of the scale. The loading value of each item exceeds 0.5, and the cumulative contribution rate of variance is 69.66% > 70%, indicating that the validity of the scale is good.

Table 3 Questionnaire scale factor analysis, convergent validity, and composite reliability.
Full size table

Convergence validity

On the premise that the confirmatory factor analysis (CFA) of each scale has a good fit, the convergence validity values of each dimension of the scale were further tested. The standardized factor load values of each measurement item in the corresponding dimension were calculated. The convergence validity and combination reliability of each dimension were calculated via the calculation formula of average variance extracted (AVE) and combination reliability (CR). Table 3 shows that the standardized load coefficients of all the factor variables were >0.5, indicating that the measured variables meet the factor requirements and pass the significance test (P < 0.05), which can be identified as having a sufficient variance explanation rate, indicating that all variables can be displayed on the same factor. The AVE values of each dimension were >0.5, and the CR values were >0.7. The synthesis shows that each dimension has good convergence validity and combination reliability. The reliability and validity test results of the factors in the four dimensions show that the KMO values of each dimension were >0.6, indicating significance at the 0.05 level. The analysis was considered valid, indicating that the scale had good reliability and accuracy. The reliability coefficient values were >0.80, indicating that each dimension of the questionnaire has credibility and that the data are real and reliable.

Discriminative validity

As shown in Table 4, the values on all diagonal lines are greater than those on the nondiagonal lines, indicating that the scale has discriminative validity. The values of the AVE of each factor after taking the square root were greater than the correlation coefficient, indicating that each factor has differential validity. In summary, all scales of the questionnaire have good structural validity, allowing for the next stage of correlation analysis.

Table 4 Correlation analysis among variables.
Full size table

In this study, Pearson correlation analysis was used to analyze the relationships between NMU, MT, and EC and farmers’ WTUCE. As shown in Table 4, there was a significant positive correlation between all scale dimensions. The correlation coefficient between new media use and farmers’ WTUCE was 0.627, and the correlation coefficient between NMU and EC was 0.625, indicating a positive correlation. Therefore, the positive correlation between NMU, EC, and farmers’ WTUCE was preliminarily verified. The above correlation analysis results only preliminarily verified the degree of correlation among the variables in this study.

Structural equation model

The structural model explains the relationship between variables through the variable covariance matrix. We prepared the structural equation model diagram shown in Fig. 2.

Fig. 2: Structural equation model.
figure 2

*Indicates significance at the 0.05 level; **indicates significance at the 0.01 level; ***indicates significance at the 0.001 level. “NMU” stands for “new media use,” “EC” stands for “ecological cognition,” and “WTUCE” stands for “willingness to use clean energy”.

Full size image

According to the model fit test results shown in Table 5, CMIN/DF = 2.899, RMSEA = 0.085 in the range of 1–3, and good in the range of 0.09. The test results of the IFL, TLI, and CFI all reach excellent levels, with values above 0.9. Therefore, the analysis results show that the model has a good fit.

Table 5 Main indicators of the model fit test.
Full size table

The path analysis results presented in Table 6 shows that NMU has a significant positive predictive effect on the WTUCE of farmers, p < 0.5, indicating that NMU can directly promote the WTUCE of farmers, verifying H1. EC has a significant positive predictive effect on the WTUCE of farmers (p < 0.5), indicating that improvements in farmers’ EC increase the WTUCE of farmers by 76.9%, which verifies Hypothesis H2. NMU has a significant positive predictive effect on EC, p < 0.5, indicating that the popularity of NMU drives farmers to have greater rural EC and realize the importance of ecology. Thus, H3 is verified. To further verify whether EC plays a mediating role in NMU and the WTUCE of farmers, it is necessary to continue the mediating effect tests.

Table 6 Results of the path relationship test.
Full size table

Intermediation effect test

This study used EC as the mediating variable to explain the relationships among NMU, EC, and farmers’ WTUCE via the following formula:

$$Y={cX}+{e}_{1}$$
(1)
$$M={aX}+{e}_{2}$$
(2)
$$Y={c}^{{\prime}}X+{bM}+{e}_{3}$$
(3)

where coefficient c in Eq. (1) is the total effect of the independent variable X on the dependent variable Y; the coefficient a in Eq. (2) is the effect of X on the intermediary variable M; the coefficient b in Eq. (3) is the effect of the intermediary variable M on Y after controlling for the influence of the independent variable X; the coefficient c’ is the direct effect of the self-variable X on the dependent variable Y after controlling for the influence of the intermediary variable M; and e1~e3 are regression residuals (Wen and Ye, 2014).

In this study, the bootstrap method was used to test the mediating effect. The bootstrap method was set for repeated sampling 5000 times, and the error correction interval was set to 95%. The final result determines whether 0 is included in the 95% confidence interval to test whether the intermediary effect exists. As shown in Table 7, EC plays a significant and complete mediating role in the relationship between NMU and farmers’ WTUCE, with an effect value of 0.588, a z value > 1.96, and a 95% confidence interval of [0.336, 0.977] without 0, which supports H5.

Table 7 Path coefficients of mediating effects.
Full size table

Moderating effect test

As shown in Table 8, MT positively affects EC, with a moderating effect value of 0.4439 and p < 0.001. The moderating effect value of the interaction term between NMU and MT is −0.0615, p < 0.05, with a 95% confidence interval [−0.1210, −0.0019], excluding 0. This finding verifies H4.

Table 8 Test results of the media–trust moderating effect.
Full size table

A simple slope test was conducted on MT, NMU, and EC. As shown in Table 9, the slopes of low MT and high MT showed a straight upward trend, indicating that, with continuous improvement in MT, awareness of EC increases constantly. Figure 3 shows the mediating effect of MT on new media use and ecological cognition. With respect to the slopes of the two lines, the slope of the low MT is greater than that of the high MT, indicating that, with increasing MT, the positive impact of NMU on EC decreases. These findings indicate that MT inversely regulates the relationship between NMU and EC. Compared with a low MT, the influence of NMU on EC is weaker under the condition of a high MT.

Table 9 Results of the simple slope test.
Full size table
Fig. 3
figure 3

Mediating effect of MT on new media use and ecological cognition.

Full size image

Results of hypothesis testing

The hypothesis test results regarding the relationships among NMU, EC, MT and the WTUCE of farmers are shown in Table 10.

Table 10 Results of hypothesis testing.
Full size table

Further analysis and discussion

Further analysis

Topographic heterogeneity analysis

We investigated the terrain in which the respondents lived and analyzed the heterogeneity from the perspective of terrain. We analyzed the difference in the effect of NMU on the WTUCE for different terrain types. The values of the coefficient c of NMU based on terrain on the WTUCE in rural areas for plains and mountainous terrains are 0.737 and 0.672, respectively. The independent variable is significantly established for the dependent variable. The mediation effect test was conducted on the premise that coefficient c is significantly established. The results can be drawn by analyzing Table 11 and Fig. 4a, b.

Table 11 Heterogeneity test from the perspective of topography.
Full size table
Fig. 4: The moderating effect of MT in the region.
figure 4

a Mountainous region. b Plains region.

Full size image

The path relationship tests indicate that NMU in plains-type areas can directly promote the WTUCE of farmers. Additionally, EC positively affects the WTUCE of farmers. NMU positively affects the level of EC, which verifies H1, H2, and H3. The mediating effect test was passed, indicating that EC in the plains areas had a mediating effect on the relationship between NMU and farmers’ WTUCE, and H5 was verified. Failure to pass the moderating effect test indicates that MT in the plains region directly affects EC, and H4 is not valid without the interaction with NMU, which reveals its impact on EC.

With respect to mountainous areas, the path relationship test was passed, indicating that NMU in mountainous areas can directly promote the WTUCE of farmers, and H1–H3 was verified. The mediating effect test was passed, indicating that EC in mountainous areas had a mediating effect on the relationship between NMU and farmers’ WTUCE, and H5 was verified. The moderating effect test indicates that MT plays a moderating role in the NMU on EC in mountainous areas, and H4 was verified.

Heterogeneity analysis on the basis of the proportion of nonagricultural income

According to the theory of development economics, developing countries have an economic development system with a dual urban–rural structure, and China is the largest developing country. Although the state has introduced policies to promote rural development, the dual structure is still a typical feature of China’s social development. At present, regarding farmers’ income sources, a large part of their income is nonagricultural income. In this study, a proportion of nonagricultural income <60% is defined as a low nonagricultural income level, and a proportion of 60% and above is defined as a high nonagricultural income level. The values of the coefficient c of NMU on the WTUCE of farmers based on the nonagricultural income difference ratio are 0.047 and 0.351, respectively. Thus, the independent variable was significantly established for the dependent variable, and the intermediary effect test was conducted on the premise that coefficient c is significantly established. The details are shown in Table 12 and Fig. 5a, b.

Table 12 Heterogeneity test from the perspective of farmers’ income.
Full size table
Fig. 5: The moderating effect of MT on nonfarming income.
figure 5

a Low nonfarming income. b High nonfarming income.

Full size image

The path relationship test was passed from the perspective of low nonagricultural income, indicating that NMU can directly promote the WTUCE of farmers from the perspective of low nonagricultural income. EC positively affects the WTUCE. NMU positively affected the level of EC, verifying H1–H3. The mediating effect test was passed, indicating that EC has a mediating effect on the relationship between NMU and farmers’ WTUCE from the perspective of low nonagricultural income, and H5 was verified. The moderating effect test indicates that MT plays a moderating role in NMU on EC from the perspective of low nonagricultural income, and H4 was verified.

From the perspective of those with a high proportion of nonagricultural income, the path relationship test was passed, indicating that NMU in mountainous areas can directly promote the WTUCE of farmers. EC positively affects the WTUCE of farmers. NMU positively affects the level of EC, which verifies H1–H3. The mediating effect test was passed, indicating that EC has a mediating effect on the relationship between NMU and farmers’ WTUCE from the perspective of high nonagricultural income, and H5 was verified. Failing the moderating effect test indicates that MT directly affects EC from the perspective of a high proportion of nonagricultural income, and it does not affect EC through interaction with NMU, meaning that H4 was not proven.

Discussion

Topography

On the basis of a review of the previous literature, Liu et al. (2017) analyzed the priorities of farmers’ demands for clean energy policies from the perspective of different terrains. They reported that farmers in hilly and mountainous areas put the policy of providing subsidies in the first place, whereas farmers in plains-type areas tend to preferentially purchase clean energy products even if the state does not provide subsidies. Moreover, plains areas are more focused on policies for technical training and on information about clean energy (Yin and Zhao, 2023). Furthermore, based on topographic heterogeneity analysis, this study revealed that heterogeneity among different topographies exists in terms of a moderating effect. MT in mountainous areas has a moderating effect on the relationship between NMU and EC, whereas MT in plains areas has no moderating effect. That is, compared with plains-type areas, the higher the degree of MT in mountainous areas is, the greater the role of NMU on EC; it plays a moderating role. Compared with that in mountainous areas, the direct effect of NMU on EC was 21.43% greater in plains areas, and the direct effect of NMU on EC was greater. This is because plains areas have more convenient access to information, a more developed economy, richer culture, and more open minds, and the direct effect of NMU on EC is stronger. On the other hand, farmers in mountainous areas consider the credibility of media and require interaction between MT and NMU if it is to have an impact on their EC (Yin et al., 2024).

Nonagricultural income proportion

On the basis of a literature review, with increasing income diversity, farmers’ consumption of commodity energy and clean energy increases, and this trend is more obvious in nonagricultural households (Su et al., 2022). Our heterogeneity analysis, which was based on the proportion of nonagricultural income, shows that the heterogeneity difference between the proportion of nonagricultural income and the level of nonagricultural income exists in terms of a moderating effect. For those with a low proportion of nonagricultural income, MT has a moderating effect on the relationship between NMU and EC, whereas MT has no moderating effect for those with a high proportion of nonagricultural income. Compared with those with a high proportion of nonagricultural income, the greater the degree of MT among those with a low proportion of nonagricultural income is, the greater the effect of NMU on EC. Compared with those with a low proportion of nonagricultural income, the direct effect of NMU on EC is 14.3% greater than that for those with a high proportion of nonagricultural income: that is, the direct effect of NMU on EC is stronger. This is because families with a high proportion of nonagricultural income work outside the home more and spend more time in cities. In cities, farmers are exposed to more advanced technologies, abundant information, and diverse thoughts and cultures (Yin et al., 2024). NMU has a stronger direct effect on EC, whereas the moderating effect of MT is not obvious. However, farmers with a low proportion of nonagricultural income consider the trustworthiness of media and require interaction between MT and NMU if it is to have an impact on their EC.

Conclusions and future prospects

Conclusion

This paper constructs a theoretical model of farmers’ WTUCE from a microscopic perspective. The results show that NMU directly affects the WTUCE of farmers. EC positively affects the WTUCE of farmers, and NMU positively affects their EC. EC has a mediating effect on the relationship between NMU and the WTUCE of farmers. Moreover, MT has a moderating effect on the relationship between NMU and EC. After these results were obtained, a heterogeneity analysis was conducted from the perspectives of terrain and nonagricultural income ratios. Compared with plains-type areas, the greater the degree of MT in mountainous areas, the greater the effect of NMU on EC; it plays a moderating role. Compared with that in mountainous areas, the direct effect of NMU on EC was 21.43% greater in plains-type areas, and the direct effect of NMU on EC was greater. Compared with those with a high proportion of nonagricultural income, the greater the degree of MT among farmers with a low proportion of nonagricultural income is, the greater the effect of NMU on their EC; it plays a moderating role. Compared with those with a low proportion of nonagricultural income, the direct effect of NMU on EC was 14.30% greater for those with a high proportion of nonagricultural income, and the direct effect of NMU on farmers’ EC was greater.

Implications

Theoretical implications

  1. a.

    From the perspective of research and content, the literature review reveals that previous studies have focused on farmers’ WTUCE in terms of income, national policies, village topography, traffic conditions, policy cognition, willingness to accept, information-receiving channels, and farmers’ cognitive level. This work enriches the theories of what affects WTUCE, especially with respect to farmers’ willingness. From the perspective of NMU, this study used farmers’ WTUCE as the dependent variable, NMU as the core explanatory variable, EC as the mediating variable between NMU and clean energy use, and MT as the moderating variable between NMU and EC. On this basis, we built a model of the factors influencing the WTUCE of farmers. The hypothesis tests were passed, confirming the validity of this model. Our predecessors have not used explanatory variables, mediating variables, or regulating variables.

  2. b.

    From the perspective of research methods, the mediating role of EC in NMU and the WTUCE of farmers was constructed; that is, NMU affects the WTUCE of farmers in rural areas by influencing their EC. We also examined the mediating role of MT in NMU and EC; that is, NMU affects farmers’ EC through the mediating role of MT. Descriptive statistical analysis tests were carried out; a structural equation model was established; hypotheses, mediation effects and adjustment effects were tested; and, finally, all the tests were passed, and the hypotheses were verified. In addition, the heterogeneity test was also applied, and we found that the test results were different for different terrain types and different levels of nonagricultural income. This suggests that different measures should be taken according to different situations. This study enriches the research methods used to investigate WTUCE.

Practical implications

  1. a.

    The perception of ease of use should be improved by optimizing NMU. First, we can build more base stations and improve infrastructure in rural areas to increase the stability of farmers’ internet access. This not only helps more farmers use the internet but also makes it more convenient for farmers to use new media and increases NMU from the source. Second, farmers can pay attention to public accounts, news media, videos, and other media related to clean energy. This will strengthen their NMU, especially in new media related to clean energy and ecological protection. In addition, full play should be given to interactive media, strong communication, and a wide range of new media characteristics. Villagers can continue to play clean energy-related radio in their villages, and related posters can be distributed, creating a WTUCE atmosphere for farmers.

  2. b.

    The fetter of trust must be broken, and the MT should be enhanced. The first way to achieve this goal is to improve the quality and authenticity of media content and ensure the reliability of information through strict news gathering and editing standards. Farmers should be provided with access to accurate information about clean energy. Second, feedback and error correction mechanisms should be established. Media organizations should establish effective feedback channels for farmers to express their views and raise questions. Finally, government supervision and guidance should be strengthened, and relevant policies and measures should be introduced to guide media organizations in improving their level of trust.

  3. c.

    Environmental protection publicity should be expanded to increase EC. Village collectives should enhance their offline publicity, the organization of ecological protection information, continuous broadcasting, and other methods. Farmers are influenced by the environment, which improves EC. Policy publicity and information release should also be strengthened. Official departments should use new media to release timely information on policies and regulations related to ecological protection and subsidy measures for clean energy. Meanwhile, feedback and suggestions from farmers can also be collected through new media platforms to provide references for the formulation of WTUCE policies.

  4. d.

    Strategies should be adapted to local conditions and vary from person to person. In plains-type areas, the government should make greater efforts to optimize NMU and increase media publicity by improving communication networks and guiding farmers to pay attention to media accounts. In mountainous areas, the government should not only optimize farmers’ NMU but also improve farmers’ MT by improving the quality and authenticity of media content. Similarly, for farmers with a high proportion of nonagricultural income, the government should pay more attention to optimizing their NMU. For farmers with a low proportion of nonagricultural income, the government should not only optimize farmers’ NMU but also improve their MT, which helps promote WTUCE.

Deficiencies and prospects

This study considered the factors influencing NMU related to farmers’ WTUCE, and this research process has had some shortcomings. In the discussion of heterogeneity, only the perspectives of terrain and the nonagricultural income share were considered, which is not sufficiently comprehensive. Therefore, the research content needs to be further expanded. Future studies can further explore the internal mechanisms of NMU on farmers’ WTUCE from the perspectives of policy formulation and implementation, energy cognition, economic income, and multiple entities. Heterogeneity analysis should also be carried out from the perspectives of farmers’ political outlook, gender, age, and education level to enrich the research results in the field of clean energy.