Global carbon emissions are exceeding 40.9 billion tons in 2023, highlighting the urgent need for clean energy transitions. China, with its commitment to carbon neutrality by 2060, faces a significant challenge in transitioning rural energy structures away from traditional, polluting sources like biomass and coal. This transition is particularly crucial in rural areas where low-income farmers often rely on these unsustainable energy sources. While national policies promote clean energy adoption in rural areas, farmer enthusiasm remains low. This study focuses on understanding the factors driving farmers' willingness to use clean energy (WTUCE) at a micro level. Previous research has explored factors like government policies, income levels, and infrastructure. However, the role of new media in shaping farmers' ecological cognition (EC) and their subsequent WTUCE has been largely overlooked. This study addresses this gap by examining how new media use (NMU), media trust (MT), and EC influence WTUCE, with a focus on the moderating effects of MT and the mediating role of EC.

Existing literature extensively examines factors influencing farmers' WTUCE, including government policies, income, education, village topography, and infrastructure. Studies consistently show that stronger policies, better infrastructure, higher income, and increased education levels correlate with greater WTUCE. Research also demonstrates the positive impact of ecological cognition (EC) on WTUCE, highlighting the importance of environmental awareness. However, few studies have explored the impact of new media use (NMU) on WTUCE. While EC is frequently cited as an influential factor, the role of media trust (MT) as a moderating variable between NMU and EC, and the mediating effect of EC between NMU and WTUCE, have not been thoroughly investigated. This study fills this knowledge gap by focusing on these under-researched relationships and their heterogeneity across different contexts.

This study employed a quantitative research design using a structured questionnaire survey. Data was collected from 263 farmers across four northern Chinese provinces (Hebei, Shandong, Shanxi, and Henan), chosen for their relatively slower pace of rural infrastructure development and economic growth. The questionnaire, distributed via both online platforms (WeChat, QQ) and in-person visits, measured four key variables: NMU, MT, EC, and WTUCE. Each variable was assessed using a five-point Likert scale. The questionnaire included items related to respondents' demographics, NMU frequency and duration, media trust (assessing credibility of information sources, media types, and information itself), ecological cognition (covering understanding of environmental issues, protection measures, and policy), and WTUCE (regarding different clean energy options). Data analysis involved descriptive statistics, reliability and validity tests (Cronbach's alpha, exploratory factor analysis (EFA), confirmatory factor analysis (CFA)), correlation analysis (Pearson correlation), and structural equation modeling (SEM) to test the proposed hypotheses. Furthermore, subgroup analyses were conducted to assess heterogeneity based on topography (plains vs. mountainous regions) and the proportion of non-agricultural income (low vs. high).

The study's SEM analysis confirmed the following hypotheses: 1) NMU positively affects WTUCE; 2) EC positively affects WTUCE; 3) NMU positively affects EC; 4) MT moderates the relationship between NMU and EC; and 5) EC mediates the relationship between NMU and WTUCE. Specifically, the direct effect of NMU on WTUCE was statistically significant. The indirect effect of NMU on WTUCE through EC was also significant, indicating a complete mediation. The moderating effect of MT on the NMU-EC relationship was confirmed, revealing that the impact of NMU on EC is stronger when MT is high. Subgroup analysis revealed significant heterogeneity across different terrains and income levels. In plains areas, NMU had a stronger direct effect on EC compared to mountainous areas, where the moderating effect of MT was more prominent. Similarly, for farmers with a low proportion of non-agricultural income, MT played a more significant moderating role in the NMU-EC relationship compared to farmers with high non-agricultural income. Numerical results from the path analysis and moderation analysis are detailed in Tables 6, 7, and 8 within the paper.

The findings underscore the critical role of new media in promoting clean energy adoption among farmers. The mediating effect of EC highlights the importance of raising environmental awareness through new media. Farmers' increased understanding of ecological issues positively influences their willingness to adopt clean energy. The moderating effect of MT suggests that trust in information sources is crucial for effective communication and behavior change. In particular, the observed heterogeneity highlights the need for context-specific strategies. In mountainous regions, building media trust is paramount to maximizing the impact of NMU on EC. Conversely, in plains regions, direct exposure to information through NMU may be more effective. Similarly, income diversification appears to influence the impact of NMU and the role of MT. These findings provide valuable insights for policymakers and program developers in designing effective clean energy promotion campaigns targeted at different rural contexts.

This study offers valuable contributions to the understanding of factors influencing farmers' WTUCE. The proposed model integrating NMU, MT, and EC offers a more comprehensive framework than previous research. The identification of significant mediation and moderation effects enhances our understanding of the underlying mechanisms. The heterogeneity analysis highlights the importance of tailoring clean energy promotion strategies to specific topographic and socioeconomic contexts. Future research could explore additional factors influencing WTUCE, such as farmers' perceived behavioral control, social norms, and specific clean energy technologies. Further research investigating the long-term impacts of NMU on sustainable energy adoption would also be beneficial.

The study's reliance on self-reported data might introduce some bias. The cross-sectional nature of the data limits the ability to establish definitive causal relationships. The sample, while geographically diverse within the study region, might not be fully representative of all rural areas in China. Further studies with longitudinal data and broader geographical coverage are recommended to enhance the generalizability of findings. The specific media platforms used by the study participants might also influence the results. Future studies could investigate the role of specific platforms and content formats in driving EC and WTUCE.