Farmers’ value perception captures their cognitive assessment of agricultural technologies and the trade-offs preceding adoption decisions (Rogers and Shoemaker, 1971; Huber et al., 2024). Technology adoption depends not only on dissemination methods but fundamentally on farmers’ perceived value of innovations. As rational economic actors, farmers exhibit inherent preference for technologies demonstrating clear utility (Gao et al., 2024). Information acquired through agricultural extension services undergoes cognitive processing and mental representation, ultimately crystallizing into value perceptions that guide decision-making (Qiao et al., 2022). Unlike traditional fertilization methods, scientific fertilization prioritizes ecological benefits and environmental protection values (Li and Ma, 2021).

Embeddedness theory asserts that individual economic actions are socio-structurally embedded within relational frameworks (Granovetter, 1973). This phenomenon manifests distinctively in China’s rural “acquaintance society,” where relational networks exert substantive influence on agricultural production decisions through multifunctional mechanisms (Gao et al., 2023). The comprehensive penetration of agricultural input distribution systems across Chinese villages has facilitated the emergence of trust-based symbiosis between farmers and private entities (e.g., agro-dealers, agribusinesses) (Manasa et al., 2024; Kumar et al., 2020; Li et al., 2019). Through longitudinal interactions, these market-oriented actors have developed “relational trust capital,” a localized social credibility that permeates their private territories. Such trust capital significantly lowers adoption barriers, enhancing farmers’ propensity to accept technologies promoted by private agricultural extension systems.

Initially, agricultural technical services primarily served as auxiliary products for agricultural inputs, functioning as key tools for private agricultural extension organizations to boost sales (Sun, 2021). However, with the development of agricultural socialized services, market players like agricultural input stores and agribusinesses have transitioned from being “single-product vendors” to “comprehensive service providers.” The emergence of innovative models like technology trusteeship has further transformed how private agricultural extension organizations participate in farmers’ production decisions (Ma et al., 2021). Private agricultural extension organizations now influence scientific fertilization adoption through two main pathways: First, they provide agronomic guidance during fertilizer sales, conveying essential information about application ratios and nutrient balance, thereby promoting proper technology adoption through information dissemination (Giulivi et al., 2022). Second, through technology trusteeship arrangements, they enable passive adoption by offering technical equipment, products, and complete fertilization management, even when farmers lack independent implementation capacity (Ma et al., 2021). This study argues that private agricultural extension organizations can shape farmers’ scientific fertilization decisions through these service-oriented approaches, either by promoting specific technologies or directly providing technical services (Figure 1).

Based on the above analysis, this study proposes the following research hypotheses:

The dataset employed in this research originates from field surveys administered to rice-growing households in Hubei Province by our research group between July and August 2022. Hubei Province was chosen as the research site based on two key considerations: First, rice stands as one of China’s three major grain crops. Hubei Province, among China’s 13 primary grain-producing regions, represents a key rice cultivation zone in the Yangtze River Basin, renowned as the ‘Jingchu Granary’. In 2021, rice cultivation in Hubei accounted for 7.60% of the nation’s total sown area and 8.85% of total production. Second, agricultural technology extension efforts in Hubei have yielded significant achievements. By 2022, the agricultural science and technology contribution rate in the province had surpassed 63%, marking an 8 percentage point increase over the previous decade. Consequently, using Hubei’s rice-growing regions as a case study holds substantial importance for elucidating the promotion mechanisms of scientific fertilization technologies and advancing the green transformation of the rice industry.

The study adopted a multi-stage random sampling method. Initially, six municipalities (Tianmen, Xiantao, Zhijiang, Yunmeng, Songzi, Zhongxiang) were selected as sampling areas based on research feasibility. Subsequently, 2–4 townships were randomly selected from each sampled county-level division. Then, 2–4 administrative villages were chosen from each selected township. Ultimately, 15–20 farming households were randomly selected as respondents in each sampled village. The survey encompassed five core modules: household demographics, agricultural operations, adoption of scientific fertilization techniques, agricultural extension services, and village characteristics. Given the predominance of middle-aged/elderly respondents with limited formal education among rice-growing households, face-to-face structured interviews were implemented. Trained postgraduate and undergraduate researchers recorded responses based on farmers’ explicit answers during field interviews. After removing questionnaires with excessive missing data or internal inconsistencies, 1,246 valid responses were retained for analysis.

Farmers’ decision-making on adopting scientific fertilization technology is not a single decision but a typical stepwise adoption process. Specifically, farmers first decide whether to adopt the technology, then adopters must further decide whether to use it according to technical specifications. Therefore, conventional binary choice models cannot meet the research requirements. Accordingly, this study employs the Heck-Probit model for estimation, with the model specified as follows:

Equations 1 and 2 represent the outcome equation and selection equation, respectively. y 1 i can only be observed when y 2 i = 1 . Here, y 1 i and y 2 i are the dependent variables, y 1 i ∗ and y 2 i ∗ represent latent variables in two groups, x 1 i and z 2 i denote vectors of explanatory variables, α and β are corresponding coefficients to be estimated, μ 1 i and μ 2 i are residual terms, with i indicating the observation sequence. Based on the sample selection mechanism, the unconditional probability model is constructed as:

In the above equations, ρ denotes the correlation coefficient, corr ( μ 1 i , μ 2 i ) = ρ . Based on Equations 3–5, the log-likelihood function is constructed as follows:

In Equation 6, n 1 represents the sample size when y 2 i = 0 ; n 2 denotes the sample size when y 2 i = 1 ; n 3 is the sample size when y 2 i = 1 ; lnL stands for the log-likelihood value; Φ ( · ) and Φ 2 ( · ) correspond to the cumulative standard normal distribution function and the cumulative bivariate normal distribution function, respectively.

This study employs the Heck-Probit model to examine the impact of private agricultural extension services on farmers’ adoption decisions regarding scientific fertilization technologies. Following Chen et al. (2020), we utilize “whether acquaintances adopt the technology” as the identification variable. This is because farmers’ adoption of green production technologies typically exhibits “herd behavior” characteristics. When more neighboring farmers adopt the technology, imitative behavior emerges, leading to adoption decisions. Numerous studies confirm the positive role of “neighborhood effects” in promoting technology adoption (Niu et al., 2022; Manda et al., 2024). However, farmers’ scientific application behaviors rely more on self-discipline than mandatory constraints. Therefore, technical training activities cannot directly influence whether farmers use technologies scientifically. The paper also constructed two simple models and found that the coefficient of influence of the identifying variables on the adoption behavior of scientific fertilizer technology by farmers was significant (p = 0.007) and the coefficient of influence on the scientific use of technology by farmers was not significant (p = 0.178). This verifies the direct impact on adoption but not implementation, justifying the identification variable selection. Table 2 reports the model estimation results.

Table 2 indicates that the Athrho value and LR test passed significance tests at 5 and 10% levels respectively, confirming sample selection bias and justifying the use of the Heck-Probit model. The Wald chi² statistic is significant at the 1% level, indicating strong model fit. Empirical results show that private agricultural extension services significantly positively influence farmers’ adoption and proper implementation of scientific fertilization techniques at the 1% level, validating Hypothesis 1. This suggests that market entities like agro-input suppliers effectively promote farmers’ adoption decisions through technical services. Therefore, policymakers should incentivize market actors to participate in fertilization initiatives and develop diversified extension systems.

Age exhibits a significant negative impact on farmers’ adoption of scientific fertilization technologies, indicating older farmers are less likely to adopt these practices. This aligns with findings that younger farmers show greater receptiveness to technological innovations and information acquisition (Wang X et al., 2023). Conversely, education level demonstrates a strong positive correlation with both adoption and proper implementation of fertilization technologies. Higher education enables farmers to better understand sustainable agricultural benefits and improves their technical literacy regarding fertilization schedules and dosage precision, leading to more scientifically sound application of adopted technologies (Wang P et al., 2024; Wang J et al., 2024).

Participation in cooperatives shows a significant positive correlation with both the adoption and scientific use of fertilization technologies within household characteristics, suggesting that cooperative involvement enhances farmers’ technical compliance. Cooperatives exert regulatory influence through unified field management and production stewardship, aligning member farmers’ practices with technical specifications to optimize adoption and application efficiency (Wang P et al., 2024). Land transfer demonstrates a significant negative impact on the scientific use of fertilization technologies. Short-term informal land leases create tenure instability, prompting farmers to prioritize low-risk conventional methods rather than fully adhering to technical protocols (Mao et al., 2021). Households with Wi-Fi connectivity demonstrate a more scientific adoption of technology. This is likely because enhanced digital access enriches farmers’ private knowledge and diminishes uncertainties in implementation.

Planting purpose shows a significant positive correlation with both adoption and scientific use of fertilization technologies. Farmers engaged in both subsistence and private production demonstrate stronger adoption and more scientific implementation, as they balance food safety concerns with profit motives while seeking to meet quality standards. The planting area has a significantly positive impact on farmers’ adoption behavior of scientific fertilization technology. Larger farm operations show greater adoption rates due to scale effects that reduce per-unit technology conversion costs and lower adoption thresholds.

This study conducted robustness tests through core explanatory variable substitution and sample restriction methods, with verification results presented in Table 3. Initially, “whether agricultural input stores or agribusinesses serve as your primary channel for obtaining technical information” was selected as the proxy variable for private agricultural extension services, coded as 0 or 1. Results demonstrate that post-substitution regression outcomes remained consistent with baseline estimations, confirming the robustness of original findings. Subsequently, subsamples with “household consumption” planting purposes were excluded for revalidation, given their documented lower demand for green technologies (Wang P et al., 2024). The obtained results aligned with baseline regression outcomes, further substantiating their robustness.

In addition, this paper also uses the endogenous switching Probit model and uses the counterfactual analysis method to estimate the average treatment effect (ATT) of farmers’ acceptance of different agricultural technology extension service organizations on farmers ‘scientific fertilization technology adoption decisions. The estimated results are shown in Table 4. The results indicate that there is a significant positive treatment effect of private agricultural extension services on both scientific fertilizer technology adoption behavior and scientific use of technology by farmers. The robustness of the benchmark regression results was further verified.

The regression results confirm that private agricultural extension services significantly enhance farmers’ scientific fertilization technology adoption decisions. To examine the underlying mechanisms, this study investigates three pathways: (1) value perception transmission, (2) symbiotic benefit generation through institutional trust, and (3) technology trusteeship provision. Following the methodological approach of Wen and Ye (2014), we analyze the impact of private agricultural extension services on farmers’ adoption decision-making for these technologies. The empirical findings are presented in Tables 5–7.

The results of Table 5 show that value perception has a significant mediating effect in the path of private agricultural technology extension services on farmers ‘scientific fertilization technology adoption behavior and technology science use behavior. This is because the private agricultural technology promotion service can transmit the information of scientific fertilization technology to farmers through technical training, thus improving the farmers ‘cognition level of the value of scientific fertilization technology, thus promoting the scientific fertilization of farmers.

Table 6 indicates that private agricultural extension services have a positive yet statistically non-significant impact on institutional trust. This suggests that while such services may enhance farmers’ trust in operating entities, the effect remains limited. A plausible explanation is that establishing farmers’ trust in business entities requires long-term communication. Higher contact frequency typically correlates with stronger trust; however, this study employs binary variables to measure private agricultural extension services, which may inadequately capture the true impact of these services on trust levels.

The results of Table 7 show that technical trusteeship has a significant mediating effect on the adoption behavior of farmers’ scientific fertilization technology by private agricultural technology extension services, and the mediating effect is not significant in the path of private agricultural technology extension services to farmers’ scientific use of technology. This may be because technical trusteeship, as a materialized service, acts more directly on the path of technology adoption behavior.