A Study on the Adoption of Digital Technologies by New Agricultural Operators Under Climate Adaptation and Sustainable Development Goals: Digital Technology Cognition, Climate Risk Perception, and Multidimensional Barriers as Moderators

Abstract

Against the backdrop of intensifying global climate change and the digital transformation of agriculture, promoting the adoption of digital technologies among new agricultural operators is a crucial pathway to enhancing agricultural climate resilience and achieving sustainable agricultural development. Based on survey data from 516 new agricultural operators in typical agricultural regions such as Northeast China, Hunan, and Hebei, this study employs Logit models, moderation effects, and heterogeneity analysis to examine the impact of digital technology cognition and climate risk on operators’ technology adoption behavior, as well as the underlying mechanisms. The findings reveal the following: First, digital technology cognition has a significant positive impact on the adoption of digital technologies, whereas climate risk perception and experiences with extreme weather significantly inhibit adoption behavior. Second, the perception of multidimensional barriers—comprising technical, economic, social, and policy obstacles—significantly moderates the positive effect of digital technology cognition on adoption behavior. Third, these effects exhibit significant heterogeneity across business scale, years in operation, and entity type. These conclusions remain valid after robustness tests and endogeneity control. This study enriches theories of agricultural technology diffusion and sustainable development from a climate resilience perspective, providing empirical evidence to promote the use of digital technologies for agricultural climate adaptation, refine differentiated extension policies, and enhance the level of agricultural sustainability.

1. Introduction

Rising global temperatures have triggered frequent extreme weather events [1], resulting not only in significant loss of life and economic damage [2], but also directly impacting the agricultural sector, with developing countries being affected to a significantly greater extent [3]. As a major agricultural nation and the largest developing country, China sees meteorological disasters account for over 70% of losses from natural disasters. From 1978 to 2024, the area of crops affected by disasters has remained at a high level, with the proportion of severely affected areas nearing 50%. This poses severe constraints on the sustainable development of agriculture and the long-term profitability of new agricultural operators, making it urgent to strengthen the climate adaptation capacity of agricultural entities as they transition toward modernization and sustainability. The Central Committee of the Communist Party of China and the State Council attach great importance to addressing agricultural climate risks. Central Document No. 1 for 2022–2025 has continuously deepened relevant deployments, progressing from theoretical research to capacity building and institutional improvement, and then to precise implementation through multi-stakeholder collaboration. This has formed a systematic approach to response, driving the transformation of agriculture from high-carbon and extensive practices toward low-carbon, resilient, green, and efficient models, aligning with the core objectives of the United Nations Sustainable Development Goals (SDGs)—“Zero Hunger,” “Climate Action,” and Sustainable Cities and Communities” [4]. At the same time, China has achieved remarkable results in building a “Digital China,” having already met the 5G and gigabit optical network construction targets for the 14th Five-Year Plan ahead of schedule. With the improvement of rural digital infrastructure, digital technology has become a key pathway for driving sustainable agricultural transformation and enhancing the sector’s resilience to climate risks. The development of digital rural areas also lays a solid foundation for the digital transformation of agriculture. Identifying the factors influencing the adoption of digital technology by new agricultural operators is of great significance for formulating relevant guidance policies and ensuring food security.

Existing domestic research on the impact of the rapid development of digital technology on the adoption of digital agricultural technologies by new agricultural operators has primarily focused on the external environment. The level of policy support and subsidy policies for technology extension services directly influence the effectiveness of technology implementation [5]; measures to improve rural digital infrastructure—such as increased internet penetration rates at the county level, seamless network coverage, and specialized technical guidance [6]—can effectively reduce adoption costs and perceived risks, thereby enhancing the willingness to adopt digital agricultural technologies; formal digital technology extension is more effective than informal extension, which can strengthen the influence of digital capabilities on farmers’ adoption of digital agricultural technologies [7]; and support measures such as technical training and economic incentives can also significantly increase the probability and level of digital technology adoption among agricultural operators [8]. Overall, existing domestic research has yielded substantial findings regarding the application value, adoption determinants, and adoption mechanisms of digital agricultural technologies. However, there remains a research gap regarding the impact and underlying mechanisms of extreme weather on the adoption of digital agricultural technologies by new types of agricultural business entities. Furthermore, there is insufficient research on the differentiated application of these technologies across different regions and among business entities of varying types and scales, and there is a lack of systematic policy package design and long-term incentive mechanisms. Further research is needed to support policy formulation.

In contrast, international research has recognized that digital agricultural technologies are a key driver in addressing the climate crisis, and related studies are showing a shift from isolated technological applications toward systematic governance. Some studies focus on the practical implementation of specific technologies, such as using multi-scale remote sensing to monitor crop water stress and establishing precision irrigation systems through IoT devices [9]. Recent studies, however, have shifted their focus to technology integration and institutional coordination, such as the “science–policy–stakeholder” tripartite linkage mechanism proposed by Ghirardelli et al. [10]. Through resilient policy design, collaborative research mechanisms, and digital platforms, these studies have effectively addressed the “mismatch in spatial scales” encountered in technology dissemination [11]. From a policy perspective, agricultural subsidy policies abroad exhibit diverse classifications, with core components including market price support, coupled subsidies linked to inputs or outputs, direct transfer payments, and support for comprehensive agricultural services. The application of digital technologies in meteorological observation and information integration provides new agricultural operators with opportunities and platforms to learn about climate crisis response measures and purchase relevant materials, thereby helping to enhance agricultural climate resilience [12]. However, foreign scholars also recognize that the empowerment brought by digital technology has given rise to new challenges. For example, algorithmic pricing power may to some extent erode farmers’ residual claims [13]; monopolies in agricultural big data platforms can easily lead to data hegemony [14]; and patent barriers for smart agricultural machinery have exacerbated developing countries’ dependence on foreign technology [15]. The case of the EU’s agricultural and environmental policy reform demonstrates that technological subsidies alone are insufficient to drive the widespread adoption of digital technologies; they must be accompanied by market access mechanisms and phased transition safeguards [16]. Overall, existing international research has largely focused on the impact of climate risks on agricultural cropping systems [17], agricultural factor inputs [18], outputs [19], and agricultural production [20]. Research on the mechanisms through which digital technologies influence the adoption of climate risk management strategies across the entire production cycle by new agricultural operators remains limited. In particular, there is a lack of empirical analysis on the heterogeneous responses of operators of different scales, and research coverage in this area is relatively sparse.

This study focuses on the adoption of digital technologies by new agricultural operators in response to climate risks. The core research questions are: How do experiences with extreme weather, perceptions of climate risks, and digital technology cognition collectively influence the decision-making process regarding the adoption of digital technologies among these operators? What are the underlying mechanisms and patterns of heterogeneity among these operators? How can we develop policies and long-term incentive mechanisms to guide the adoption of digital technologies in the context of climate risk scenarios?

Based on the research questions outlined above, this study focuses on the factors influencing the adoption of digital technologies by new agricultural operators in response to climate risks. It aims to promote the adoption of relevant digital technologies, enhance their capacity to respond to climate risks and build resilience, identify key influencing factors, alleviate the mismatch between technology supply and operational capabilities, and refine theoretical frameworks while proposing policy recommendations. Current academic research on the diffusion of digital agricultural technologies tends to emphasize economic factors and macro-policy effects, while studies on the adoption of such technologies by new agricultural operators in the context of climate risk response remain limited. In particular, there is a lack of research that integrates technological cognition, climate risk expectations, experiences with extreme weather, and multiple socio-economic and cultural barriers into a unified analytical framework to systematically reveal direct and moderating effects. By constructing an integrated analytical framework, this study examines three key dimensions—perception of climate risks, experiences with extreme weather, and levels of digital technology awareness—while simultaneously investigating the inhibitory effects of risk factors and the transmission pathways of various barriers. This approach effectively addresses the shortcomings of existing research, such as narrow perspectives and insufficient mechanism analysis, and further reveals the extent of influence exerted by various factors on adoption decisions. It can expand the theoretical framework of agricultural climate adaptation, laying a theoretical foundation for targeted policy implementation and future research. The findings provide scientific evidence to support governments in designing differentiated support policies, assist agricultural producers in optimizing technology application, and enable R&D professionals to advance technological iteration. Consequently, this research enhances agricultural climate resilience, reduces disaster losses, promotes the deep integration of digital technology with agriculture, and contributes to the construction of a climate-smart agricultural system and the process of agricultural modernization. The findings provide a scientific basis for governments to formulate policies on the digital transformation of sustainable agriculture, for new agricultural operators to optimize green and low-carbon production, and for technology developers to advance tailored innovations. They will help enhance agricultural climate resilience, reduce ecological and economic losses, and promote the deep integration of digital technologies with sustainable agriculture, thereby contributing to the development of climate-smart agricultural systems and the process of agricultural modernization.

2. Conceptual Definitions and Research Hypotheses

The adoption of digital agricultural technologies is not merely a technical issue, but rather a managerial decision involving expected returns, resource endowments, risk exposure, institutional support, learning capacity, and social interactions [21,22]. Early research on agricultural innovation adoption primarily explained farmers’ technology adoption behavior in terms of expected returns, farm size, human capital, credit availability, access to information, extension services, and risk tolerance [22]. This research tradition has corrected the simplistic view that “technologically advanced innovations will inevitably be adopted,” emphasizing that adoption decisions depend on the economic viability, operational suitability, and institutional environment of technologies [22,23]. With the development of precision agriculture and smart agriculture, the research focus has gradually shifted from traditional agricultural technologies to digital agricultural technologies [24,25]. Unlike traditional technologies, digital agricultural technologies are characterized by data-intensity, knowledge-intensity, and service dependency, integrating elements such as sensors, remote sensing, artificial intelligence, big data analytics, decision support systems, cloud platforms, and mobile applications [26,27].

Rogers’ diffusion theory suggests that technology diffusion is influenced by perceived characteristics such as relative advantage, compatibility, complexity, trialability, and observability [28]. However, empirical studies in agricultural economics show that the diffusion of agricultural innovations is uneven, largely due to heterogeneity in expected returns, production conditions, and resource endowments across farmers [29]. Farmers are not passive adopters but rational decision-makers operating under resource constraints and limited opportunities [30,31]. Incomplete information, uncertainty, credit constraints, and heterogeneity are central to understanding agricultural technology adoption [32]. Adoption or non-adoption depends not only on the technical advantages of innovations but also on their economic feasibility and institutional context [23,33]. Importantly, non-adoption should not be interpreted as backwardness, as farmers may rationally reject technologies when costs, risks, or incompatibility outweigh benefits [34,35]. Therefore, the adoption of digital technologies should be understood as a constrained and conditional choice rather than an automatic response to technological progress [36].

With the incorporation of behavioral decision-making theories into agricultural technology adoption research, frameworks such as the Theory of Planned Behavior, the Technology Acceptance Model, and the Unified Theory of Acceptance and Use of Technology have been widely used to explain how perceived usefulness, perceived ease of use, subjective norms, and perceived behavioral control influence adoption intentions [37,38]. While these models provide important insights into cognitive mechanisms, the relationship between cognition and actual adoption remains context-dependent and may be moderated by external constraints such as risk and resource availability [39].

Some studies have focused on climate adaptation behaviors and agricultural resilience. Smit and Skinner classified agricultural adaptation strategies, noting that adaptation measures can encompass multiple aspects, including technical, managerial, financial, and institutional dimensions [40]. Farmers’ adaptation to climate change is influenced by a combination of biophysical exposure and socioeconomic conditions [41,42]. Furthermore, adaptation behaviors are subject to social constraints, as they are shaped by values, institutions, knowledge, and resources [43]. Farmers’ choice of adaptation strategies is influenced by their educational level, access to climate information, availability of extension services, and household characteristics [44]. Farmers’ perceptions of climate play a significant role in adaptation responses [45]. Farmers’ beliefs and concerns regarding climate change affect their level of support for adaptation and mitigation measures [46]. Furthermore, agricultural stakeholders’ views on climate change are not uniform but are shaped by social, political, and personal experiences [47]. There is a gap between climate information and farmers’ use of decision-support tools; awareness does not necessarily lead to action [48,49]. While there is an association between climate beliefs and adaptation strategies, this association is moderated by factors such as water availability, farm context, and risk attitudes [50]. Adaptive capacity depends not only on objective resources but also on human cognition, perceived efficacy, and risk assessment [51]. Climate risks influence the demand for digital agricultural technologies, but the impact of perceptions of future risks and past experiences on technology adoption remains to be verified.

Research focusing on precision agriculture, smart agriculture, and digital agriculture indicates that, compared to traditional agricultural technologies, digital agricultural technologies rely more heavily on data, knowledge, and related services. The adoption of precision agriculture is influenced by factors such as farm size, profitability, farmer characteristics, information sources, and external support [52]. Technological characteristics, economic factors, and institutional environments also collectively shape decision-making regarding precision agriculture adoption [25]. Smart agriculture systems involve socio-ethical challenges, including farmer autonomy, advisory service relationships, and data governance issues [53]. These studies indicate that the adoption of digital agricultural technologies is not only related to perceived usefulness but is also inextricably linked to infrastructure, trust, service systems, and governance arrangements. Research focusing on agricultural innovation systems and socio-technical support reveals that innovation capacity depends on interactions among actors rather than isolated technology transfer [54,55]. Innovation brokers and adaptive management can support transformation within agricultural systems [56]. Rural innovation relies on communication, social learning, and negotiation [57]. As farmers’ needs become increasingly diverse and knowledge-intensive, services such as agricultural extension face new challenges and must be adapted to local conditions to achieve optimal alignment [58,59]. Without adequate infrastructure and economic support, even operators with strong digital literacy may be unable or unwilling to adopt these technologies; however, there remains a lack of systematic examination of how the barriers faced by new agricultural operators influence final adoption.

In the context of precision agriculture, farmers’ decisions regarding technology adoption are influenced by factors such as expected profitability, farm size, technological complexity, compatibility with existing production systems, and the availability of advisory services [23,24]. Empirical evidence further suggests that the adoption of precision agriculture technologies varies across regions and farmer groups, reflecting differences in production systems and resource endowments [23]. The emergence of smart agriculture technologies has further highlighted the importance of data, connectivity, skills, and service support, implying that technology adoption depends not only on individual willingness but also on the broader agricultural innovation system [27]. From the perspective of digital agriculture, issues such as data ownership, platform dependency, institutional coordination, and farmer engagement have become key constraints affecting technology application [27,60]. Research on agricultural mobile applications and digital decision-support tools indicates that while perceived utility and ease of use are important, actual adoption depends on practical needs, functionality, and production contexts [37]. When climate risks are incorporated into analytical frameworks, farmers’ adoption of digital technologies should be understood as a decision-making process shaped by perceived benefits, perceived risks, resource constraints, institutional support, and behavioral cognition, and should account for heterogeneity [39,61].

The adoption of digital agricultural technologies lies at the intersection of technology acceptance, climate adaptation, risk response, and innovation systems. Existing research has confirmed that factors such as digital technology cognition, perceived usefulness, access to information, risk perception, and institutional support have a significant impact on adoption behavior. However, existing research has not yet examined the role of digital technology cognition in the context of climate risks, has paid insufficient attention to situations where climate risks may trigger defensive rather than innovative responses, and has failed to simultaneously integrate the two risk pathways of climate risk perception and experiences of extreme weather events. While digital agriculture research emphasizes barriers such as digital literacy and infrastructure, it provides insufficient explanation of how multidimensional barriers—including technological, economic, social, and policy barriers—influence the “cognition-to-adoption” transformation mechanism. To address these gaps, this study proposes that the adoption of digital technologies should be understood within the joint constraints of digital technology cognition, climate risk perception, and the perception of multidimensional barriers. Rather than treating digital adoption as a simple extension of technological adoption models, this study conceptualizes it as a decision-making behavior jointly shaped by cognitive levels, climate shocks, and the perception of barriers. The relationship between cognition and adoption is not static but varies across contexts. Under climate risks, adoption decisions must pass through multiple filters of climate risk perception, perceived feasibility, and past experiences. By integrating digital technology cognition, climate risk perception, multidimensional barriers, and climate adaptation and sustainable development goals, this study reveals an integrated pathway for research on farmer adoption. Consequently, it enhances the explanatory power of technology acceptance models in climate contexts and supplements and expands the theoretical framework of agricultural technology adoption.

2.1. Digital Technology Cognition and Digital Technology Adoption Behavior

The issue of digital technology cognition has long been a key area of research for scholars both domestically and internationally. This study primarily investigates the level of digital technology cognition among new agricultural operators, defining digital technology cognition as their understanding of the value and operational aspects of the digital technologies they are familiar with [62].

Digital technology cognition influences the adoption of digital technologies. The impact of digital technology cognition on technology adoption is rooted in the core proposition of the Technology Acceptance Model (TAM): an individual’s adoption decision depends on their subjective evaluation of the technology [63]. Among these, “perceived usefulness” and “perceived ease of use” are widely recognized as core variables influencing the willingness of new agricultural operators to adopt digital technologies [64], and they exert a significant influence on technology adoption [65].

The Technology Acceptance Model (TAM) explains and predicts individuals’ acceptance and usage of new technologies through two core variables: perceived usefulness and perceived ease of use. Perceived usefulness refers to the extent to which an individual believes that using a new technology can improve work efficiency, while perceived ease of use refers to the ease with which an individual believes a new technology can be used [66]. Based on the Technology Acceptance Model (TAM) and its core variables, when new agricultural operators cognitively determine that digital technology offers significant advantages in helping them improve production efficiency—that is, when they recognize that digital technology has high utility value—they will develop positive adoption expectations. When these operators recognize that the operational barriers or learning costs associated with digital technology are within an acceptable range, their reluctance toward the new technology will decrease, making them more likely to adopt digital technology. Therefore, when the digital technology cognition of new agricultural operators is at a higher level, their likelihood of adopting digital technology increases; that is, digital technology cognition has a positive impact on technology adoption behavior.

Digital technology cognition is a key factor influencing the adoption of digital technologies by new agricultural operators [67]. Digital technology cognition has a significant positive impact on both the adoption decisions and the degree of adoption among new agricultural operators [68], playing a positive role in the technology adoption process [69]. However, when new agricultural operators face limitations in their digital technology cognition, they tend to be less likely to adopt digital technologies [70]. Focusing on the perception of benefits within digital technology cognition, new agricultural operators with higher levels of digital technology cognition exhibit a higher degree of adoption of digital technologies [71].

Based on the above analysis, this study proposes Hypothesis 1.

H1. 

Digital technology cognition of new agricultural operators will promote the adoption of digital technologies.

2.2. The Impact of Climate Risk Perception and Experiences with Extreme Weather on the Adoption of Digital Technologies

Climate risk perception and the number of extreme weather events experienced will influence the adoption of digital technologies. In recent years, climate risks have become frequent, and the increasing number of experienced extreme climate events poses a significant threat to agricultural production by new agricultural operators, creating uncertainty regarding their expectations for future agricultural yields. This study uses prospect theory to explain the relationship between climate risk perception and the adoption of digital technologies, and uses threat rigidity theory to explain the relationship between the number of extreme weather events experienced and the adoption of digital technologies.

According to prospect theory [72], when faced with uncertainty, people base their decisions not on their final wealth level, but on gains and losses relative to a specific reference point; furthermore, people are more sensitive to losses than to gains, a phenomenon known as loss aversion. In this study, the psychological reference point for new agricultural operators is their current production model, specifically the status quo of not adopting digital technologies. New agricultural operators are influenced by the high upfront investment costs of new technologies and the uncertainty of expected returns [73], meaning they face both certain risks and potential risks. According to prospect theory, prior losses intensify the pain of subsequent losses; therefore, the presence of certain losses and the potential for future probabilistic losses will inhibit the adoption behavior of new agricultural operators. Issues such as rising temperatures and precipitation imbalances caused by climate change increase agricultural production risks and uncertainty [74], which may result in losses for new agricultural operators; however, these losses are characterized by time lags and uncertainty. Compared to the certain losses associated with technology adoption, new agricultural operators assign a lower decision-making weight to climate-related risks; that is, they place greater emphasis on certain losses and consequently refuse to adopt digital technologies.

According to the theory of threat rigidity, when organizations or individuals face threats of operational difficulties arising from internal or external environments, they develop an extremely strong fear of failure [75], enter a state of extreme risk aversion, and lose their ability to manage [76]. In this study, the increasing number of extreme weather events experienced by new agricultural operators indicates that they face a broader range of threats, which in turn heightens their perception of uncertainty. According to the theory of threat rigidity, new agricultural operators tend to adopt conservative strategies when facing threats and reject the exploration of new technologies.

Climate risk perception and the increasing frequency of extreme weather events expose new agricultural operators to risks, making agricultural production more vulnerable to natural disasters and subjecting their production revenues to greater uncertainty [77]. Furthermore, these new agricultural operators’ risk aversion inhibits their technology adoption behavior [78]. At the same time, new technologies expose these new agricultural operators to varying degrees of uncertainty or risk, thereby inhibiting their adoption of digital technologies [79]. Under conditions of ambiguity—that is, uncertainty—risk and ambiguity preferences significantly influence technology selection; new agricultural operators who are averse to ambiguity tend to choose safer options [80]. That is, when faced with the threat of uncertainty and risk, new agricultural operators are more inclined to adopt conservative strategies rather than adopt new technologies.

Based on the above analysis, this paper proposes the following hypotheses:

H2. 

High climate risk perceptions for the future among new agricultural operators will inhibit the adoption of digital technologies.

H3. 

When faced with an increasing number of extreme weather events, new agricultural operators tend to adopt conservative strategies rather than embrace new technologies.

2.3. The Moderating Role of Perceived Multidimensional Barriers

Multidimensional barriers mitigate the impact of digital technology cognition on the adoption of digital technologies. As mentioned earlier, high digital technology cognition promotes the adoption of digital technologies, while the presence of multidimensional barriers inhibits such adoption. In this paper, multidimensional barriers include technological, economic, social, and policy barriers. Numerous studies in agricultural economics, both domestically and internationally, have examined the impact of barrier factors on technology adoption. It is currently widely recognized that the barrier factors affecting the adoption of digital technologies span multiple dimensions. Cui, L., et al. categorized these factors into five major categories: socioeconomic, agroecological, technological, institutional, and psychological and behavioral [81]. Jaron Porciello et al. categorized influencing factors into four dimensions: performance expectancy, effort expectancy, social influence, and facilitating conditions, with a corresponding number of influencing factors under each dimension [31]. Following a literature review, Ulucak S Z et al. proposed a constraint typology model encompassing five major barrier domains: economic, technical, political, sociocultural, and environmental. Since climate risk and environmental factors form the primary research context of this study, this paper adopts technical, economic, social, and policy barriers as the main barrier factors. Based on the literature by Ulucak S Z et al., the four types of barriers are explained as follows: technical barriers refer to a lack of technical information, unmet training needs for agricultural professionals, a shortage of (supporting) equipment, inadequate infrastructure, supply chain shortages, uncertainty regarding future impacts, high perceived risk, and technical complexity; economic barriers include insufficient farm size, lack of credit, and high costs associated with technology, credit, labor, land, and operations, as well as long payback periods and high economic uncertainty; social barriers include prejudice against new technologies, negative influences from social networks and peers, negative media coverage, traditional beliefs and practices, insufficient knowledge and negative perceptions among new agricultural operators, misinformation about new technologies, and strong resistance to changing habits; policy barriers include insufficient incentives, lack of subsidies, regulatory restrictions, bureaucratic procedures, lack of policy support, and property rights issues [82].

According to the Theory of Planned Behavior, all factors that may influence behavior indirectly affect behavioral performance through behavioral intention. Behavioral intention, in turn, is influenced by three variables: behavioral attitude—an individual’s evaluation of how much they like or dislike the consequences of performing a particular behavior; subjective norm—the social pressure an individual feels when deciding whether to perform a particular behavior, reflecting the influence of others or groups on the individual’s behavioral decision; and perceived behavioral control—the individual’s perception of how easy or difficult it is to perform a particular behavior [83]. The intention to adopt technology among new agricultural operators—that is, behavioral intention—depends not only on their positive perception of the technology itself (i.e., behavioral attitude) but is also influenced by their assessment of the ease or difficulty of the adoption process (i.e., perceived behavioral control). The perceptions of technological, economic, social, and policy barriers examined in this study essentially conceptualize negative perceived behavioral control in this specific context. According to the Theory of Planned Behavior, when new agricultural operators face high implementation resistance—that is, when they encounter barriers—their behavioral intention to adopt digital technology is significantly inhibited, even if they hold a positive attitude toward its adoption. Therefore, we introduce the perception of barriers as a key moderating variable to explain why some new agricultural operators with high levels of digital technology cognition fail to adopt digital technology after perceiving significant external barriers.

In terms of technological barriers, successful adoption requires new agricultural operators to trust the technology, undergo technical training, and find the technology easy to use [84]. That is, technological simplicity promotes adoption, while technological complexity inhibits it. Furthermore, when technical barriers exist, new agricultural operators may develop a fear of difficulty, leading to a decline in their perceived trust in digital technologies, which in turn suppresses the positive impact of their digital technology cognition on the adoption of digital technologies [85,86]. Second, regarding economic barriers, when such barriers exist—particularly high technology costs—they can dampen the positive impact of digital technology cognition on the adoption of digital technologies, thereby reducing the likelihood that new agricultural operators will adopt these technologies [87]. Next, regarding social barriers, when new agricultural operators receive inaccurate or one-sided information about a technology—that is, when they receive false information about a new technology—they tend not to adopt that technology [88]. At the same time, social networks may also lead to class-based disparities in information access, hindering the distribution of information resources and thereby inhibiting technology adoption [89]. Finally, regarding policy barriers, when funding for agricultural technology extension is reduced and the extension system is characterized by compartmentalization across sectors, departments, and regions [90], that is, a lack of policy support and systemic flaws can inhibit technology adoption. When the government implements regulatory systems—that is, when it employs command-and-control measures—the probability that new agricultural operators will not adopt digital technologies also increases [91].

Based on the above analysis, this paper proposes the following hypothesis:

H4. 

Multidimensional barriers perception will moderate the positive effect of digital technology cognition on digital technology adoption among new agricultural operators.

Based on the theoretical analysis presented above, this paper constructs an analytical framework, as shown in Figure 1.

3. Empirical Design

3.1. Data Sources

The data in this paper is derived from a special questionnaire survey on “Climate Risks and Digital Technology Adoption” conducted by the research team in 2025 targeting new agricultural operators. The survey focused on regions such as Northeast China, Hunan, and Hebei. The Northeast region is primarily characterized by large-scale grain cultivation and is significantly affected by low temperatures and frost damage; Hunan is primarily a rice-growing region, significantly affected by high temperatures and flooding; Hebei features a mix of grain, vegetable, and fruit production, and is notably affected by drought and pests and diseases. These regions encompass China’s major agricultural production models in terms of agricultural characteristics and exhibit diverse types of climate risks, providing an excellent research setting for examining the adoption of digital agricultural technologies by new agricultural operators in response to climate risks. New agricultural operators primarily include specialized large-scale farmers, family farms, farmers’ cooperatives, agricultural enterprises, and other entities. Compared to traditional smallholder farmers, they possess distinct advantages in terms of technology adoption and risk management. The survey employed a questionnaire-based approach: multiple typical agricultural regions, including Northeast China, Hunan, and Hebei, were selected. Questionnaires were distributed randomly within each region, and all collected questionnaires were consolidated. Valid questionnaires were then selected based on their content for data analysis.

The survey questionnaire primarily consists of the following modules: (1) Basic characteristics of new agricultural operators; (2) Digital technology cognition, risk perception, and expectations; (3) Technology adoption; (4) New agricultural operators’ perceptions of influencing factors. A total of 586 questionnaires were distributed for this survey. After rigorous verification of the returned questionnaires and the exclusion of invalid responses due to missing key information or logical inconsistencies, 516 valid questionnaires were ultimately obtained for empirical analysis, resulting in a response rate of 88.1%.

The data descriptions and descriptive statistics for the dependent variable, core explanatory variables, mechanism variables, and control variables are presented in

Table 1. Variable Declaration and Assignment.

Variable Classes	Variable Name	Definitions and Assignments
Dependent variable	Adopt	Have you used digital technology in agricultural production? Yes = 1, No = 0
Core explanatory variable	Tech_know	Digital technology cognition. What is your level of understanding of digital agricultural technology: 1 = not at all familiar, 2 = not very familiar, 3 = relatively familiar, 4 = very familiar.
	Risk_expect	Your expectations for climate risks over the next 3 years
	risk_extreme	Number of extreme climate events experienced in the past decade
Moderating variable	Diff_tech	Impact level of technical application difficulties: No impact = 1, Minimal impact = 2, Moderate impact = 3, Significant impact = 4, Very significant impact = 5
	Diff_econ	Impact level of economic and resource-related difficulties: assigned values as above (1–5)
	Diff_soc	Impact level of social and cultural difficulties: Valuation as above (1–5)
	Diff_policy	Impact level of policy and institutional difficulties: assigned values as above (1–5)
Control variables (individual characteristics)	Gender	Your gender: Male = 1, Female = 2
	Age	Your age: 18–25 years = 1, 26–40 years = 2, 41–50 years = 3, 51–60 years = 4
	Edu	Your education level: Elementary school or below = 1, Junior high school = 2, High school/vocational school = 3, College/University or above = 4
(Operational characteristics)	Scale	Land management area: less than 50 mu = 1, 50–200 mu = 2, more than 200 mu = 3
	Type	Specialized large-scale farmers = 1, Family farm = 2, Farmers’ cooperative = 3, Agricultural enterprise = 4, Others = 5
	Year	Establishment years of the business entity: 1–3 years = 1, 4–6 years = 2, 7–10 years = 3, over 10 years = 4
	Income_ration	Proportion of agricultural income in total household income: <20% = 1, 20–40% = 2, 40–60% = 3, 60–80% = 4, >80% = 5

Note: Digital technology cognition is based on Davis (1993) [63]; technical, economic, social, and policy barriers are based on Ulucak et al. (2026) [82]; the remaining variables were collected directly based on the actual survey findings.

,

Table 2. Descriptive statistics for continuous variables.

Variable Name	Mean	Standard Deviation	Minimum	Maximum
Tech_know	2.10	0.87	1	4
Risk_expect	2.82	1.42	1	5
risk_extreme	2.39	1.39	0	6
Diff_tech	3.83	1.05	1	5
Diff_econ	3.90	1.00	1	5
Diff_soc	3.71	1.03	1	5
Diff_policy	3.81	1.03	1	5
Age	2.86	0.93	1	4
Edu	2.54	0.93	1	5
Scale	1.67	0.77	1	3
Year	2.07	1.20	1	4
Income_ration	2.88	1.49	1	5

and

Table 3. Descriptive statistics for categorical variables.

Variable Name	Category	Frequency	Percentage
adopt	Adopted (=1)	103	19.9%
	Not adopted (=0)	413	80.0%
gender	Male (=1)	349	67.6%
	Female (=2)	167	32.3%
type	Specialized large-scale farmers (=1)	103	19.9%
	Family farm (=2)	214	41.4%
	Farmers’ cooperative (=3)	76	14.7%
	Agricultural enterprise (=4)	62	12.0%
	Others (=5)	61	11.8%

.

3.2. Variable Selection and Description

3.2.1. Dependent Variable

This study uses the adoption of digital technologies—specifically, whether new agricultural operators adopt digital technologies—as the dependent variable. If an operator adopts digital technologies, it is assigned a value of 1; if not, it is assigned a value of 0. According to the survey data, approximately 20% of the operators in the sample have adopted digital technologies. This figure is broadly consistent with the agricultural production informatization rate cited in the White Paper on the Development of Digital Rural Areas (2024) [92] and is therefore fairly representative.

3.2.2. Core Explanatory Variables

This study identifies operator type, climate risk perception, and the number of extreme weather events experienced as the core explanatory variables.

• Digital Technology Cognition (Tech_know): Level of familiarity with digital technologies among new agricultural operators. This variable takes values ranging from 1 to 4, corresponding to “not at all familiar” through “very familiar” (see Table 1), with higher values indicating a higher level of familiarity. This variable focuses on the operators’ subjective understanding of the functions and practical value of digital technologies.
• Climate Risk Perception (Risk_expect): New agricultural operators’ predictions regarding future climate risks. Values range from “significantly reduced” to “significantly increased” (see Table 1); higher scores indicate stronger climate risk perception, reflecting new agricultural operators’ subjective assessments of uncertainty in agricultural production and influencing their demand for risk management through technological means;
• Experience with Extreme Weather (risk_extreme): The number of distinct types of extreme weather events actually experienced by new agricultural operators over the past 10 years. This metric counts the total number of different types of events experienced (e.g., drought, flooding, hail, prolonged heatwaves, prolonged cold spells, etc.). A higher value indicates a greater variety of extreme weather events experienced, which influences the new agricultural operators’ willingness to adopt digital technologies.

3.2.3. Moderating Variables

This study identifies four types of barriers faced by new agricultural operators during the adoption of digital technologies as moderating variables, specifically including technical barriers, economic barriers, social barriers, and policy barriers (see Table 1). These barriers are important moderating factors influencing the final adoption decisions of new agricultural operators. According to theoretical analysis, the strength of the influence of digital technology cognition on adoption behavior varies depending on the perceived severity of these barriers; that is, the perception of barriers acts as a moderator between digital technology cognition and adoption behavior.

• Technical barriers (diff_tech): This reflects the new agricultural operators’ perceptions regarding the operational complexity and learning difficulty of digital technologies.
• Economic barriers (diff_econ): This reflects the new agricultural operators’ perceptions regarding the high investment costs and uncertain returns associated with digital technologies.
• Social barriers (diff_soc): This reflects new agricultural operators’ perceptions regarding the lack of technical training and insufficient role modeling from surrounding groups.
• Policy Barriers (diff_policy): This reflects new agricultural operators’ perceptions regarding the lack of government subsidies and insufficient policy support.

3.2.4. Control Variables

To accurately identify the net effect of the core explanatory variables on digital technology adoption behavior, this study selects a series of factors that may influence adoption decisions as control variables. These primarily cover individual and operational characteristics, including: gender (gender), with a value of 1 for male and 2 for female; age (age), representing the actual age group of the new agricultural operators; education level (edu), representing the actual educational attainment of the new agricultural operators; operational scale (scale), representing the range of land area managed by the new agricultural operators; operator type (type), representing the actual registered type of the new agricultural operators; years in operation (year), representing the actual duration of the new agricultural operators’ operation; Income Ratio (income_ratio), representing the proportion of agricultural income in total household income (see Table 1).

3.2.5. Description of Variable Measures

The measurement of digital technology cognition draws on the relevant dimensions from the Acceptance Model [63]; the measurement of multidimensional barriers draws on the constraint typology framework [82]. The remaining variables were collected directly based on the actual survey context. The specific measurement methods and scoring for each variable are presented in Table 1.

3.3. Model Specification

3.3.1. Baseline Regression Model

To empirically examine the impact of digital technology cognition, climate risk perception, and the number of extreme weather events experienced on the adoption of digital technology by new agricultural operators, this study constructs the following baseline Logit model:

$$\mathrm{A}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{T}\mathrm{e}\mathrm{c}\mathrm{h}\_\mathrm{k}\mathrm{n}\mathrm{o}{\mathrm{w}}_{\mathrm{i}}+{\mathsf{\alpha }}_4\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}{\mathrm{s}}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{r}}+{\mathsf{ϵ}}_{\mathrm{i}}$$

(1)

$$\mathrm{A}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_2\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}\_\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{c}{\mathrm{t}}_{\mathrm{i}}+{\mathsf{\alpha }}_4\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}{\mathrm{s}}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{r}}+{\mathsf{ϵ}}_{\mathrm{i}}$$

(2)

$$\mathrm{A}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_3\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}10_{\mathrm{i}}+{\mathsf{\alpha }}_4\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}{\mathrm{s}}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{r}}+{\mathsf{ϵ}}_{\mathrm{i}}$$

(3)

Here, $\mathrm{A}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}}$ is the binary dependent variable, indicating whether the i-th new agricultural operators adopts digital technology; the core explanatory variables include: digital technology cognition ($\mathrm{T}\mathrm{e}\mathrm{c}\mathrm{h}\_\mathrm{k}\mathrm{n}\mathrm{o}{\mathrm{w}}_{\mathrm{i}}$), climate risk perception ($\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}\_\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{c}{\mathrm{t}}_{\mathrm{i}}$), and the number of extreme climate events experienced ($\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}10_{\mathrm{i}}$); $\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}{\mathrm{s}}_{\mathrm{i}}$ represents a set of control variables; ${\mathsf{\mu }}_{\mathrm{r}}$ and ${\mathsf{ϵ}}_{\mathrm{i}}$ denote the regional dummy variable and the random error term, respectively. The estimated coefficients ${\mathsf{\alpha }}_1$, ${\mathsf{\alpha }}_2$, and ${\mathsf{\alpha }}_3$ are the key parameters of interest. If they are significantly positive, this indicates that higher levels of digital technology cognition, stronger climate risk perception, or a greater number of experienced extreme weather events are more likely to promote the adoption of digital technologies.

Based on this, the three core explanatory variables were simultaneously included in the regression model to comprehensively examine their combined effects:

$$\mathrm{A}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1\mathrm{T}\mathrm{e}\mathrm{c}\mathrm{h}\_\mathrm{k}\mathrm{n}\mathrm{o}{\mathrm{w}}_{\mathrm{i}}+{\mathsf{\alpha }}_2\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}\_\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{e}\mathrm{c}{\mathrm{t}}_{\mathrm{i}}+{\mathsf{\alpha }}_3\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}10_{\mathrm{i}}+{\mathsf{\alpha }}_4\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}{\mathrm{s}}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{r}}+{\mathsf{ϵ}}_{\mathrm{i}}$$

(4)

In Model (4), α₁, α₂, and α₃ represent the net effects of each factor on digital technology adoption behavior, controlling for the other two core variables.

To mitigate endogeneity issues arising from potential sample selection bias, this study employs the propensity score matching method. Since the propensity score matching method requires the treatment variable to be a binary variable, we convert the digital technology cognition variable into a binary variable to construct the treatment and control groups: the sample is split based on the median value, with samples scoring at or above the median assigned to the treatment group and those below the median assigned to the control group. We then estimate the treatment effects of each variable on digital technology adoption behavior. First, a Logit model is used to estimate the conditional probability of a sample being assigned to the treatment group, i.e., the propensity score:

$$\mathrm{P}\left({\mathrm{X}}_{\mathrm{i}}\right)=\mathrm{P}\mathrm{r}(\mathrm{T}\mathrm{r}\mathrm{e}\mathrm{a}{\mathrm{t}}_{\mathrm{i}}=1|{\mathrm{X}}_{\mathrm{i}})$$

(5)

Based on the propensity scores, each sample in the treatment group is matched with the control group sample with the closest score. Finally, the difference in the outcome variable between the two matched groups is compared to obtain the average treatment effect:

$$\mathrm{A}\mathrm{T}\mathrm{T}=\mathrm{E}\left[\mathrm{a}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}1}\right|\mathrm{T}\mathrm{r}\mathrm{e}\mathrm{a}{\mathrm{t}}_{\mathrm{i}}=1]-\mathrm{E}[\mathrm{a}\mathrm{d}\mathrm{o}\mathrm{p}{\mathrm{t}}_{\mathrm{i}0}|\mathrm{T}\mathrm{r}\mathrm{e}\mathrm{a}{\mathrm{t}}_{\mathrm{i}}=1]$$

(6)

The model for estimating the propensity score is:

$$\mathrm{P}(\mathrm{T}\mathrm{e}\mathrm{c}\mathrm{h}\_\mathrm{h}\mathrm{i}\mathrm{g}{\mathrm{h}}_{\mathrm{i}}=1|{\mathrm{X}}_{\mathrm{i}})=\mathrm{F}({\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{g}\mathrm{e}\mathrm{n}\mathrm{d}\mathrm{e}{\mathrm{r}}_{\mathrm{i}}+{\mathsf{\beta }}_2\mathrm{a}\mathrm{g}{\mathrm{e}}_{\mathrm{i}}+{\mathsf{\beta }}_3\mathrm{y}\mathrm{e}\mathrm{a}{\mathrm{r}}_{\mathrm{i}}+{\mathsf{\beta }}_4\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{e}\_\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}{\mathrm{o}}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{r}}$$

(7)

We employ multiple matching methods—nearest neighbor matching, radius matching, and kernel matching—to conduct balance tests. If the treatment and control groups pass the balance tests after matching, and the estimated average treatment effect aligns with the conclusions of the baseline regression, this indicates that the core findings are robust.

3.3.2. Moderation Effect Testing Model

To further examine the moderating role of various barriers in the process by which digital technology cognition influences digital technology adoption behavior, this study treats the four types of barriers—technological, economic, social, and policy—as moderating variables ${\mathrm{M}}_{\mathrm{i}}$ to test whether they moderate the relationship between digital technology cognition and adoption behavior. To test this moderation effect, the following model is constructed:

$$\begin{array}{c}{\mathrm{Adopt}}_{\mathrm{i}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1{\mathrm{Tech}\_\mathrm{know}}_{\mathrm{i}}+{\mathsf{\gamma }}_2{\mathrm{Risk}\_\mathrm{expect}}_{\mathrm{i}}+{\mathsf{\gamma }}_3{\mathrm{Risk}10}_{\mathrm{i}}+{\mathsf{\gamma }}_4{\mathrm{M}}_{\mathrm{i}}\\ +{\mathsf{\gamma }}_5({\mathrm{Tech}\_\mathrm{know}}_{\mathrm{i}}\times {\mathrm{M}}_{\mathrm{i}})+{\mathsf{\gamma }}_6{\mathrm{Controls}}_{\mathrm{i}}+{\mathsf{\mu }}_{\mathrm{r}}+{\mathsf{\epsilon }}_{\mathrm{i}}\end{array}$$

(8)

In this model, M_i represents four types of barriers: technological, economic, social, and policy-related; Tech_know_i × M_i denotes the interaction term between the two, while the other variables are the same as in Equation (4). The coefficient γ₅ is the key moderator parameter of interest; if it is significantly positive (or negative), it indicates that the perception of barriers exerts a positive (or negative) moderating effect on the relationship between digital technology cognition and adoption behavior.

3.3.3. Heterogeneity Analysis

To examine differences in the impact of core variables on the adoption of digital technologies across varying business scales, years in operation, and business entity types, this study conducted a heterogeneity analysis, and the results were consistent with the baseline model.

The control variables included gender (Genderi), age (Agei), education level (Edui), business scale (Scalei), years in operation (Yeari), business entity type (Typei), and income ratio (Income_ratioi). To assess heterogeneity, the total sample was divided into different subsamples based on business scale (less than 50 mu, 50–200 mu, and more than 200 mu), years in operation (1–3 years, 4–6 years, 7–10 years, and more than 10 years), and business entity type (specialized large-scale farmers, family farms, farmers’ cooperatives, agricultural enterprises, and others).

4. Analysis of Empirical Results

4.1. Baseline Regression Results

To systematically examine the decision-making logic underlying the adoption of digital technologies by new agricultural operators to address climate risks, this study constructs an empirical model with digital technology adoption as the core dependent variable.

Table 4. Regression Results.

Variables	(1)	(2)	(3)	(4)	(5)
tech_know	0.770 ***			0.819 ***	0.117 ***
	(0.151)			(0.153)	(0.020)
risk_expect		−0.257 ***		−0.289 ***	−0.041 ***
		(0.085)		(0.091)	(0.013)
risk_extreme			−0.254 ***	−0.208 **	−0.030 **
			(0.087)	(0.091)	(0.013)
gender	0.087	0.084	0.091	0.161	0.023
	(0.252)	(0.247)	(0.246)	(0.261)	(0.037)
age	−0.014	−0.037	−0.055	−0.076	(0.011)
	(0.132)	(0.127)	(0.128)	(0.134)	(0.019)
edu	−0.006	0.163	0.147	0.042	0.006
	(0.136)	(0.132)	(0.131)	(0.143)	(0.020)
scale	0.016	0.125	0.133	0.023	0.003
	(0.168)	(0.164)	(0.165)	(0.170)	(0.024)
year	−0.007	0.044	0.034	0.038	0.005
	(0.095)	(0.094)	(0.092)	(0.100)	(0.014)
income_ratio	−0.089	−0.043	−0.056	−0.035	(0.005)
	(0.084)	(0.083)	(0.080)	(0.086)	(0.012)
Constant	−2.957 ***	−1.305	−1.293	−2.127 **
	(0.827)	(0.810)	(0.830)	(0.899)
Observed value	516	516	516	516
Pseudo-R2	0.068	0.024	0.021	0.103
AIC	497.85	519.53	520.85	482.60
BIC	530.97	553.50	554.82	525.06

Note: ** and *** indicate significance at the 5% and 1% levels, respectively; the numbers in parentheses are robust standard errors.

presents the results of the baseline regression analyzing the effects of digital technology cognition, climate risk perception, and experiences with extreme weather on the adoption of digital technologies by new agricultural operators. To mitigate omitted variable bias and ensure the reliability of the estimates, this study compared model fit using the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Ultimately, the baseline model—which includes digital technology cognition, climate risk perception, extreme weather risks, and all control variables, as shown in Table 4 (Column 4)—was constructed and reported. A Logit model was employed to report marginal effects, thereby identifying the impact of key factors on farmers’ technology adoption behavior. Empirical results show that the digital technology cognition variable is significantly positive at the 1% statistical level, with an average marginal effect of 0.117. Holding other factors constant, a one-unit increase in farmers’ digital technology cognition significantly raises the probability of technology adoption by 11.7 percentage points. Digital technology cognition drives the adoption of digital technologies by new agricultural operators, thereby validating Hypothesis H1; Climate risk perception is significantly negative at the 1% level, with a marginal effect of −0.041; for every one-unit increase in climate risk perception, the probability of adoption decreases by 4.1 percentage points; the factor of experience with extreme climate events is significantly negative at the 5% level, with a marginal effect of −0.030; for every one-unit increase, the probability of adoption decreases by 3.0 percentage points. New agricultural operators with higher climate risk perceptions and a greater number of extreme climate events have a lower probability of adopting digital technologies and are more inclined to maintain traditional production models; Hypotheses H2 and H3 are thus supported.

4.2. Robustness Tests

To further validate the robustness of the baseline findings, we conducted robustness tests to rule out the interference of differences in estimation methods and potential model biases on the core conclusions, thereby ensuring the rigor and reliability of the research findings.

Table 5. Results of Robustness Tests.

Variables	(1)	(2)	(3)
tech_know	0.819 ***	0.469 ***	0.796 ***
	(0.153)	(0.081)	(0.150)
risk_expect	−0.289 ***	−0.168 ***	−0.281 ***
	(0.091)	(0.050)	(0.089)
risk_extreme	−0.208 **	−0.122 **	−0.200 **
	(0.091)	(0.054)	(0.089)
gender	0.161	0.087	0.158
	(0.261)	(0.149)	(0.256)
age	−0.076	−0.039	−0.075
	(0.134)	(0.078)	(0.132)
edu	0.042	0.033	0.041
	(0.143)	(0.077)	(0.140)
scale	0.023	0.012	0.024
	(0.170)	(0.097)	(0.167)
year	0.038	0.025	0.039
	(0.100)	(0.058)	(0.098)
income_ratio	−0.035	−0.027	−0.035
	(0.086)	(0.053)	(0.084)
Constant	−2.127 **	−1.249 ***	−2.081 **
	(0.899)	(0.465)	(0.882)
Observed value	516	516	516

Note: ** and *** indicate significance at the 5% and 1% levels, respectively; the numbers in parentheses are robust standard errors.

presents the results of the robustness tests. The regression results in Column (2) show that after testing the replacement of the baseline Logit model with a Probit model, the signs, significance levels, and mechanisms of action of the core explanatory variables remained consistent between the two models. Digital technology cognition is significantly positive at the 1% level, indicating that higher levels of digital technology cognition consistently promote the adoption of digital technology; climate risk perception is significantly negative at the 1% level, and experiences of extreme climate risk are significantly negative at the 5% level. These findings are fully consistent with the negative inhibitory effect of risk perception observed in the baseline regression, confirming that the constraining effect of climate risk on adoption behavior is not due to model specification. To further mitigate potential bias issues, this study employs a Relogit model for regression analysis, correcting for rare events or potential estimation biases to enhance the robustness of the results. As shown in Table 5 (Column 3), the positive facilitating effect of digital technology cognition on technology adoption and the negative inhibitory effects of climate risk perception and extreme climate experiences have not undergone any substantial changes. The magnitude and significance of the coefficients for the core variables are highly consistent with the baseline results, further confirming the stability of the baseline conclusions. The above fully demonstrates the robustness of the baseline conclusions, and H1, H2, and H3 have been supported.

4.3. Endogeneity Tests

The baseline regression results presented earlier reveal that digital technology cognition significantly promotes the adoption of digital technologies; however, this association may be confounded by potential endogeneity issues, leading to biased model estimates. To effectively mitigate endogeneity issues arising from sample self-selection and further ensure the accuracy of the model’s matched estimates, this study employs propensity score matching to overcome the effects of selection bias.

In this study, digital technology cognition was designated as the treatment group, and balance tests were conducted on the propensity scores of the treatment and control groups to assess the conditions for the common support domain. As shown in the results of the balance test in

Table 6. Test results of the balance test.

Match Status	Pseudo-R2	LR Statistic	p-Value	Average Standardized Deviation (%)	B-Value (%)
Before matching	0.027	16.91	0.010	12.6	40.6
After matching	0.003	2.81	0.833	4.2	12.7

Note: The LR statistic corresponds to the goodness-of-fit test result for logistic regression.

, pseudo-R² decreased from 0.027 to 0.003, the LR statistic decreased to 2.81 with a p-value of 0.833, the average standardized deviation decreased from 12.6% to 4.2%, and the B-value decreased from 40.6% to 12.7%. Furthermore, the standardized deviations of all covariates fell below 10%, all meeting the criteria for the balance test. The matching effectively balanced the covariate characteristics between the treatment and control groups, significantly mitigating sample selection bias. The balance assumption holds, and the matching results are reliable. Figure 2 shows the kernel density function plots for the treatment group and the control group before and after matching. It can be seen that the overlap between the two groups significantly increased after matching, indicating that the propensity score intervals for the high-tech digital technology cognition group and the low-knowledge cognition group have largely overlapped, thereby satisfying the common support domain hypothesis.

As shown in

Table 7. Estimation Results of the PSM Model.

Matching Methods	Average Treatment Effect on the Treated (ATT)	Standard Error	t-Statistic
Nearest Neighbor Matching	0.102 **	0.044	2.34
Radius matching	0.110 ***	0.035	3.17
Kernel Matching	0.110 ***	0.035	3.18

Note: ** and *** indicate significance at the 5% and 1% levels, respectively.

, results from the three matching methods—nearest neighbor matching, radius matching, and kernel matching—indicate that technology cognition has a significant positive impact on farmers’ adoption of digital technologies. If farmers with high digital technology cognition do not possess high digital technology cognition, their probability of adopting digital technologies is relatively low, whereas farmers with high digital technology cognition exhibit a significantly higher probability of adopting digital technologies. The average treatment effects for the three matching strategies—nearest neighbor matching, radius matching, and kernel matching—are all highly significant. This indicates that high digital technology cognition significantly increases the probability of farmers adopting digital technologies, further validating the reliability of the previous analysis. Based on the results in Table 6 and Table 7, the conclusions of the baseline regression are confirmed: the promotional effect of digital technology cognition on adoption behavior remains robust, and H1 is supported.

4.4. Testing the Moderating Effects of Multidimensional Perception of Barriers

The baseline regression and endogeneity treatment discussed earlier both confirmed the critical role of digital technology cognition and experiences with climate risks in driving the adoption of digital technologies by new agricultural operators. To further explore whether multidimensional barriers influence the adoption of digital technologies by these operators, we introduced a composite measure of farmers’ perceived barriers as a moderating variable and constructed interaction terms to test its moderating effect. This study quantifies the level of farmers’ perceived barriers across four dimensions—technical, economic, social, and policy barriers—and derives a composite score for farmers’ perceived barriers using principal component analysis. The results of the reliability and validity tests show that the overall Cronbach’s alpha coefficient of the scale is 0.8722, indicating good internal consistency; the overall KMO value was 0.8135, and Bartlett’s sphericity test statistic was 1076.85 with p < 0.001. The dimensions exhibit strong inter-correlation, indicating a well-designed scale structure capable of effectively measuring the comprehensive barrier perception characteristics of new agricultural operators. After centering the variables, a moderation effect analysis was conducted;

Table 8. Results of the moderation effect analysis.

Variables	(1)	(2)
c_tech_know	0.819 ***	0.801 ***
	(0.153)	(0.155)
c_diff_composite		0.039
		(0.071)
Tech × diff		−0.140 *
		(0.074)
risk_expect	−0.289 ***	−0.295 ***
	(0.091)	(0.091)
risk_extreme	−0.208 **	−0.214 **
	(0.091)	(0.093)
gender	0.161	0.201
	(0.261)	(0.265)
age	−0.076	−0.079
	(0.134)	(0.133)
edu	0.042	0.058
	(0.143)	(0.143)
scale	0.023	0.038
	(0.170)	(0.172)
year	0.038	0.029
	(0.100)	(0.100)
income_ratio	−0.035	−0.046
	(0.086)	(0.085)
Constant	−2.127 **	−2.161 **
	(0.899)	(0.891)
Observed value	516	516
Pseudo-R2	0.103	0.110

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively; the numbers in parentheses are robust standard errors.

presents the results of the baseline regression and the moderation effect regression.

Based on the estimated results in Columns (1) and (2) of Table 8, it can be seen that after adding the comprehensive barrier perception score and its interaction term with digital technology cognition to the baseline model, the coefficient for digital technology cognition remains significantly positive at the 1% level, while the interaction term coefficient is −0.140 and is significantly negative at the 10% level. This indicates that farmers’ comprehensive barrier perception score plays a significant negative moderating role between digital technology cognition and digital technology adoption. Thus, H4 has been verified.

4.5. Heterogeneity Analysis

To gain a deeper understanding of the mechanisms through which digital technology cognition, risk perception, and experience influence the adoption of digital technologies, this study further examines the varying effects of these factors across different groups. This helps identify key target groups and provides empirical evidence for the development of targeted policies.

4.5.1. Analysis of Heterogeneity Based on Years Since Establishment

As can be seen from

Table 9. Results of the heterogeneity analysis by years since establishment.

Variables	(1)	(2)	(3)	(4)
	1–3 Years	4–6 Years	7–10 Years	10+ Years
tech_know	0.916 ***	0.478	3.733 ***	1.446 ***
	(0.258)	(0.321)	(0.940)	(0.492)
risk_expect	−0.328 **	−0.952 ***	0.181	0.255
	(0.137)	(0.272)	(0.473)	(0.247)
risk_extreme	−0.217	−0.202	−0.678	−0.573 **
	(0.155)	(0.238)	(0.607)	(0.249)
Constant	−2.740 **	2.888	−5.244	−4.129 **
	(1.192)	(1.915)	(4.490)	(1.911)
control variable	Yes	Yes	Yes	Yes
Observed value	237	117	47	104
Pseudo-R2	0.171	0.305	0.495	0.375

Note: ** and *** indicate significance at the 5% and 1% levels, respectively; the numbers in parentheses are robust standard errors.

, digital technology cognition was significantly positive across the 1–3 year, 7–10 year, and over 10 year groups, with all results reaching the 1% significance level. This indicates that digital technology cognition has a consistent, positive effect on adoption behavior for most entities, regardless of their years in operation. In the 4–6 year group, the coefficient for digital technology cognition was positive but not significant; entities in this stage may be in the process of adjusting or stabilizing their business models, and their adoption behavior may not be entirely driven by digital technology cognition. Regarding climate risk perception, risk perception was significantly negative in both the 1–3 year and 4–6 year groups, with the negative impact being stronger in the 4–6 year group. This suggests that entities with shorter operating histories or those in the growth stage are more sensitive to risk perception. When they perceive higher risks, they are more likely to delay or abandon adoption. In the 7–10 year and over 10 year groups, climate risk perception did not exhibit a significant inhibitory effect; as business experience accumulates, entities’ ability to identify, control, and tolerate risks has strengthened. Further analysis of intergroup comparisons reveals that heterogeneity in the duration of operation dimension is primarily reflected in differences in the role of climate risk perception. Specifically, the marginal effect of risk expectations in the 4–6 year group was significantly lower than that in the 1–3 year group, while the marginal effect in the over 10 years group was significantly higher than that in the 4–6 year group. This indicates that entities at different stages of development exhibit distinct responses to risk expectations. Entities with a shorter operating history have limited business experience and are more likely to adopt cautious strategies when facing risks; conversely, entities with a longer operating history have accumulated extensive experience over time, resulting in stronger risk-bearing capacity and response capabilities. Consequently, the inhibitory effect of risk expectations on their decision-making behavior is relatively weaker.

4.5.2. Analysis of Heterogeneity Based on Farm Size

As shown in

Table 10. Results of the Analysis of Heterogeneity in Business Scale.

Variables	(1)	(2)	(3)
	Under 50 mu	50–200 mu	Over 200 mu
tech_know	0.903 ***	0.070	1.891 ***
	(0.235)	(0.379)	(0.490)
risk_expect	−0.289 **	0.047	−0.678 **
	(0.137)	(0.218)	(0.302)
risk_extreme	−0.395 **	−0.383	−0.108
	(0.154)	(0.277)	(0.261)
Constant	−1.766 *	0.591	−5.351 ***
	(0.919)	(1.523)	(1.758)
control variable	Yes	Yes	Yes
Observed value	269	150	95
Pseudo-R2	0.176	0.208	0.388

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively; the numbers in parentheses are robust standard errors.

, the coefficient for digital technology cognition is significantly positive in both the groups with farm sizes under 50 mu and over 200 mu, and is significant at the 1% level in both cases. This indicates that digital technology cognition significantly promotes adoption behavior among both small-scale and large-scale farmers. Although the coefficient for digital technology cognition in the 50–200 mu group is positive, it does not reach the significance level, suggesting that medium-scale new agricultural operators—lacking the resources of large-scale entities and the flexibility of small-scale ones—are only marginally influenced by digital technology cognition in their adoption decisions. Regarding risk expectations, risk expectations are significantly negative in both the under-50 mu and over-200 mu groups; when risk expectations are high, the adoption probability of small-scale and large-scale new agricultural operators decreases significantly. Extreme risk experience was significantly negative only in the group with less than 50 mu, indicating that the adoption behavior of new agricultural operators is more easily constrained by risk-averse tendencies. Further analysis of the intergroup comparison results reveals significant differences across groups in terms of operational scale. Specifically, the marginal effect of digital technology cognition in the 50–200 mu group was significantly lower than that in the group with less than 50 mu, while the marginal effect in the group with more than 200 mu was significantly higher than in both the group with less than 50 mu and the 50–200 mu group. The promotional effect of digital technology cognition on adoption behavior does not follow a simple linear relationship with farm size but exhibits differentiated characteristics across different scale stages. Small-scale entities, likely due to limited resources, are more sensitive to and adaptable to improvements in digital technology cognition; large-scale entities, owing to higher levels of operational specialization and stronger demand for technology application, find that improvements in digital technology cognition are more easily translated into actual adoption behavior; medium-scale entities, which may be in a phase of expansion and adjustment, face constraints on their adoption behavior from multiple factors such as capital, labor, and organizational management, resulting in a relatively less pronounced effect of digital technology cognition.

4.5.3. Heterogeneity Analysis Based on Operator Type

Given that the sample sizes for some operator types are relatively small, this study employs the Firth Logit model for estimation in the heterogeneity test of operator types to address small-sample bias or the problem of complete prediction. Furthermore, average marginal effects are calculated to enhance the interpretability of the results. As shown in

Table 11. Results of Heterogeneity Analysis by Operator Type.

Variables	(1)	(2)	(3)	(4)	(5)
	Specialized Large-Scale Farmers	Family Farm	Farmers’ Cooperative	Agricultural Companies	Other
tech_know	0.405	0.995 ***	1.624 ***	1.343 **	0.592
	(0.265)	(0.248)	(0.524)	(0.683)	(0.462)
risk_expect	−0.386 *	−0.581 ***	0.071	−0.203	0.401
	(0.209)	(0.153)	(0.255)	(0.303)	(0.278)
risk_extreme	−0.150	−0.101	0.215	−0.338	−0.335
	(0.211)	(0.158)	(0.259)	(0.330)	(0.317)
Constant	−0.676	0.046	−6.153 **	−3.852 *	−3.395
	(1.897)	(1.345)	(2.651)	(2.321)	(2.599)
control variable	Yes	Yes	Yes	Yes	Yes
Observed value	103	214	76	62	61
Pseudo-R2	0.337	0.337	0.558	0.583	0.570

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively; the numbers in parentheses are robust standard errors.

, digital technology cognition exerts a significant positive influence across all three groups: family farms, farmers’ cooperatives, and agricultural enterprises. Specifically, the effect of digital technology cognition is significant at the 1% level for both the family farm and farmer cooperative groups, while it is significant at the 5% level for the agricultural enterprise group. Based on the average marginal effect results, digital technology cognition has the strongest promotional effect on farmer cooperatives, with an average marginal effect of 0.221. This indicates that for every one-unit increase in digital technology cognition, the probability of adoption behavior among farmer cooperatives increases by an average of 22.1 percentage points. The average marginal effect of digital technology cognition for agricultural enterprises was 0.174, while that for family farms was 0.134. This suggests that digital technology cognition has a more pronounced promotional effect on entities with higher levels of organization or greater operational specialization. However, the promotional effect of digital technology cognition on the adoption behavior of specialized large-scale farmers and other operator types is relatively unstable. Regarding climate risk perception, the risk expectation coefficients for both specialized large-scale farmers and family farms were significantly negative. In contrast, the impact of risk perception on farmers’ cooperatives, agricultural enterprises, and other types of entities did not pass the significance test. Farmers’ cooperatives and agricultural enterprises possess strong organizational coordination, resource integration, and risk-sharing capabilities, which can mitigate the inhibitory effect of risk expectations on adoption behavior to a certain extent. Although specialized large-scale farmers and family farms have a certain scale of operations and autonomous decision-making capacity, they are relatively limited in terms of financial strength, capacity for technological trial and error, and risk-diversification mechanisms; therefore, they are more susceptible to the influence of climate risk perception. In terms of the operator type, the impact of extreme climate experiences on adoption behavior is limited.

5. Conclusions

5.1. Policy Implications

Based on the above findings, the study offers the following policy implications:

First, attention should be paid to the pivotal role that digital technology cognition plays in driving the adoption of digital technologies by new agricultural operators, and tiered and categorized strategies for enhancing digital technology cognition should be implemented for different groups. The government must not only leverage platforms such as the agricultural technology extension system and digital rural service stations to establish a training system that integrates online and offline approaches, thereby expanding the reach of digital technology cognition; it must also provide basic hands-on training for small-scale and other entities with low risk resilience, while offering specialized, in-depth training for large-scale, highly organized entities. This will precisely enhance the digital technology cognition and application capabilities of different operator types, thereby stimulating their willingness to adopt digital technologies to address climate risks.

Second, recognizing the impact of multi-dimensional barriers, efforts must be made from multiple angles to comprehensively reduce adoption barriers, overcome these obstacles, and mitigate the negative effects of moderating variables. Efforts should be made to strengthen rural digital infrastructure, improve after-sales maintenance systems, and resolve technical barriers such as operational difficulties and network reliability. Concurrently, the application process for digital technology subsidies should be streamlined, and subsidy standards and coverage expanded to alleviate economic barriers. A digital technology exchange platform for new agricultural operators should be established to break the path dependence on traditional production experience and resolve social barriers. Additionally, specialized policies integrating climate risk and digital agriculture should be introduced, and a policy consultation and support mechanism should be established to eliminate policy barriers.

Third, given the significant heterogeneity across different groups in terms of digital technology cognition and climate risk perception—factors that influence the adoption of digital technologies—targeted measures are needed to address the specific adoption challenges faced by different groups. The government should strengthen climate risk protection for small-scale and other entities with low risk resilience, expand the coverage of climate index insurance, and provide small-scale subsidies and credit support to mitigate the conservative decision-making resulting from their risk aversion. It should encourage large-scale entities to establish digital agriculture demonstration bases to serve as models and lead the way, helping them transform their experiences with extreme weather into motivation for adoption. At the same time, the government should support farmers’ cooperatives and agricultural enterprises in pooling resources to build collective digital technology application platforms, leveraging the advantages of organization to reduce the costs and risks of adoption for individual entities.

Efforts should be made to mitigate the inhibitory effects of climate risk-related factors and stimulate adoption. A routine climate risk early warning and public education system should be established to disseminate climate trend forecasts and digital technology response plans through authoritative channels, helping new agricultural operators rationally assess risks. Successful cases of digital technology application following extreme weather events should be summarized and promoted to strengthen demonstration and leadership, guiding operators to transform risk experiences into motivation for adoption.

Moreover, supporting mechanisms should be improved to amplify scale and organizational effects. Encourage all entities to continuously improve their organizational governance structures and enhance their capabilities for risk diversification and risk perception transformation. Further refine the institutional mechanisms related to the transfer of farmland management rights and large-scale land management, promote the standardized operation of farmland transaction markets, help agricultural operators reduce transaction costs, and guide them to appropriately expand their operational scale through land transfers. This will fully unleash the catalytic effect of economies of scale on digital technology adoption and enhance the overall digital capabilities of new agricultural operators in addressing climate risks.

5.2. Discussion and Insights

Digital technology is a key pathway to promoting sustainable agricultural development, enhancing climate resilience, and achieving a green and low-carbon transition. Against the backdrop of climate risks, this study deepens and expands the theoretical understanding of digital technology adoption among new agricultural operators. It systematically integrates empirical findings into the theoretical frameworks of technology acceptance theory, behavioral decision-making theory, agricultural technology diffusion theory, and climate adaptation theory. This not only provides empirical insights into digital technology adoption but also represents a further extension of agricultural technology adoption theory in the context of climate risks. Existing research has largely focused on the influence of individual characteristics of new agricultural operators on the adoption of digital technologies, while relatively neglecting the roles of multiple factors—such as digital technology cognition, climate risk perception, and past experiences—and their underlying transmission mechanisms. Furthermore, there has been insufficient exploration of the differentiated response characteristics among different operator types. Based on survey data from 516 new agricultural operators in climate risk-prone regions such as Northeast China, Hunan, and Hebei, this study employs Logit models, propensity score matching, moderation effect tests, and heterogeneity analysis to investigate the impact and mechanisms through which digital technology cognition, climate risk perception, and experiences with extreme climate risks influence the adoption of digital technology by new agricultural operators. The following research conclusions were drawn:

First, there is significant directional heterogeneity in the effects of digital technology cognition, climate risk perception, and past climate experiences on the adoption of digital technologies by new agricultural operators. Digital technology cognition significantly and positively promotes the adoption of digital technologies, while climate risk perception and past experiences of extreme climate events both exert significant negative inhibitory effects, highlighting the role of core factors as key drivers. The significant positive impact of digital technology cognition on adoption behavior reinforces and extends the explanatory power of the Technology Acceptance Model (TAM). The TAM posits that an individual’s willingness to adopt technology depends primarily on perceived usefulness and perceived ease of use [63,64]. The empirical results of this study indicate that when new agricultural operators fully recognize the practical value and advantages of digital technology, their probability of adoption increases significantly. This conclusion is consistent with research on agricultural technology adoption, which posits that farmers’ behavior is essentially a rational decision based on expected benefits, feasibility, and resource constraints [22,23]. Therefore, this study further reinforces the theoretical proposition that “cognition is a prerequisite for adoption”; the negative effects of climate risk perception and experiences with extreme weather provide an important supplement and refinement to climate adaptation theory. Consequently, this study extends Prospect Theory [72] and Threat Rigidity Theory to the field of digital agricultural technology adoption. Under sustained stress, individuals tend to adopt conservative strategies, are more sensitive to certain losses, and place lower weight on future gains; both perceived future climate risks and past experiences of extreme weather exert significant inhibitory effects.

Second, the perception of multidimensional barriers plays a significant negative moderating role between digital technology cognition and the adoption of digital technologies, thereby expanding the Theory of Planned Behavior (TPB) and the theory of agricultural innovation systems. As evidenced by the results of the moderation analysis, the combined perception of four types of barriers—technological, economic, social, and policy—significantly weakens the positive influence of digital technology cognition on adoption behavior. When new agricultural operators perceive high adoption barriers, the positive effect of digital technology cognition is significantly weakened, making it difficult to effectively translate this cognition into actual adoption behavior. Technology adoption depends on the institutional environment, service systems, and multi-stakeholder interactions [54,56,57]. The perception of multidimensional barriers plays a crucial moderating role in determining whether digital technology cognition translates into adoption decisions.

Third, there is significant group heterogeneity in the effects of digital technology cognition, climate risk perception, and experiences with extreme weather on digital technology adoption. While digital technology cognition is a key driver of adoption behavior, its impact varies depending on the characteristics of the actors and their operational conditions. Climate risk factors primarily exert a inhibitory effect, which is more pronounced among actors with lower risk tolerance or lower levels of organization. This study extends research on the influence of resource endowments and organizational capabilities on adoption to the context of climate risk and digital transformation [22,24].

By validating the positive role of digital technology cognition in agricultural digital transformation under climate risk conditions, this study reinforces the applicability of Technology Acceptance Theory in the agricultural sector; supplements climate adaptation theory by demonstrating that climate risk perception and experiences of extreme weather may trigger conservative behavioral responses; and, by introducing the perception of multidimensional barriers as a moderating mechanism, revises and enriches the theory of agricultural technology diffusion, explaining why higher levels of digital technology cognition do not fully translate into actual adoption behavior. These theoretical contributions provide a more integrated analytical framework for understanding the digital technology adoption behavior of new agricultural operators under the goals of climate adaptation and sustainable agricultural development.

Combined with the empirical findings of this study, these results can provide theoretical support and practical guidance for new agricultural operators in making production decisions, managing climate risks, and undergoing digital transformation. The study clarifies the core driving value of digital technology cognition in agricultural digital transformation. It helps new agricultural operators objectively recognize the practical value of digital technologies in meteorological early warning, disaster prevention and control, and ensuring stable and secure production. It elucidates the intrinsic role of digital applications in mitigating climate shocks and enhancing production resilience, thereby providing directional guidance for optimizing their own production models. The study verifies the inhibitory effect of experiences with extreme weather and climate risk perception on digital adoption behavior. It objectively demonstrates that excessive risk aversion and negative disaster experiences can easily lead to rigid production decision-making. This helps new agricultural operators confront cognitive biases regarding climate risks, strike a reasonable balance between risk prevention and technological innovation, and reduce the developmental limitations caused by conservative business thinking. The conclusion regarding the negative moderating effect of multidimensional barrier perception clearly demonstrates that practical constraints—such as technical limitations, investment costs, reliance on traditional approaches, and insufficient policy alignment—can weaken the efficiency of translating awareness into actual action. This enables new agricultural operators to objectively assess the practical bottlenecks in digital transformation based on their own operational conditions and reasonably plan the pace of technological investment and implementation strategies. Based on the results of heterogeneity analysis across different business scales, key features, and degrees of organization, this approach can be tailored to the specific characteristics of various operator types, providing a reference for the differentiated selection of digital tools and the adoption of either streamlined or systematic digital agriculture models. The advantages in technology adoption offered by highly organized collaborative models also provide a practical basis for new agricultural operators to deepen cooperative ties, share digitalization costs through collective resources, and jointly build climate risk response systems, thereby continuously enhancing overall operational stability and climate resilience.

This paper contributes to the broader discussion on technology application and climate change adaptation in three key ways: First, it expands the research context by extending the adoption of digital technologies from conventional production settings to climate risk response scenarios, thereby strengthening research on the link between technology application and enhanced climate resilience and enriching the discussion on technical pathways for agricultural climate adaptation. Second, it deepens the explanation of underlying mechanisms by integrating risk perception, experiences with extreme weather, digital technology cognition, and external environmental factors. This reveals the adoption mechanisms under the interaction of multiple factors, addresses the shortcomings of single-perspective studies, and provides new evidence from a climate perspective for theories of technology diffusion. Third, it reinforces a practice-oriented approach by identifying the heterogeneous response patterns of new agricultural operators of different scales and types, proposing actionable policy combinations and incentive mechanisms, and offering Chinese experience and a reference framework for developing countries to leverage digital technologies to enhance agricultural climate resilience.

Although this paper reveals the roles of digital technology cognition, climate risk perception, and past experiences in the adoption of digital technologies by new agricultural operators, as well as the impact of perceived multidimensional barriers, it still has limitations. The research data captures only the behavioral characteristics of new agricultural operators at a specific point in time, failing to depict the dynamic evolution of how perceptions of digital technology, experiences with extreme climate risks, and climate risk perceptions influence digital technology adoption. Consequently, it is difficult to analyze the sustained impact of long-term climate risk changes and improvements in digital technology cognition on adoption behavior. To address these research limitations and in light of trends in agricultural digital technology development and climate risk management, future studies could employ panel data to conduct longitudinal research. This would allow for long-term monitoring of changes in how digital technology cognition, experiences with extreme climate risks, and climate risk perception influence the adoption behavior of new agricultural operators. Such research would analyze the dynamic evolution patterns and long-term impact mechanisms, thereby revealing the long-term interactive relationship between climate risks and digital technology adoption.