The Impact of Farmers’ Cognition and Attitudes on the Adoption of Conservation Tillage: An Empirical Study Based on the Lishu Model

Abstract

Promoting conservation tillage (CT) is central to advancing ecological transformation in China’s agricultural sector under the “Dual Carbon” strategy. However, despite strong institutional support, the adoption of CT technologies—especially the “Lishu Model” integrating no-till and straw mulching—remains uneven. This study applies the Cognition–Attitude–Behavior (CAB) framework to examine the behavioral mechanisms underlying CT adoption, drawing on a 2024 household survey in Lishu County, Jilin Province. Descriptive and regression analyses reveal that, while cognitive awareness is positively associated with adoption, only attitudinal alignment has a significant predictive effect. The findings identify a distinct group of “informed non-adopters”, underscoring that knowledge alone does not ensure behavioral uptake. These results highlight the importance of motivational alignment over information provision in shaping sustainable technology adoption and suggest the need for more psychologically grounded policy interventions.

1. Introduction

Promoting environmentally sustainable agricultural practices is a key pillar of China’s “Dual Carbon” strategy and broader goals for ecological modernization in rural areas. Among these practices, conservation tillage (CT) has gained prominence for its demonstrated benefits in reducing soil erosion, improving water retention, and enhancing long-term soil productivity. In this context, the “Lishu Model”—which integrates mechanized no-till farming with straw mulching—has been formally adopted under the Black Soil Protection Engineering Implementation Plan (2021–2025) as a flagship initiative in Northeast China’s core grain-producing region.

Yet, despite extensive institutional support and financial subsidies, the actual uptake of conservation tillage among farmers remains inconsistent. This gap between policy design and behavioral response is not unique to China. Globally, many sustainable agricultural practices face under-adoption due to a combination of technical uncertainty, unclear economic returns, and perceived risks [1,2]. While conventional adoption models emphasize rational cost–benefit calculations, they often overlook the psychological, social, and contextual factors that influence farmers’ decision-making.

Considering these limitations, an increasing number of scholars have adopted behavioral frameworks to better explain farmers’ decision-making processes. One such approach is the Cognition–Attitude–Behavior (CAB) model, which originates from social psychology and is conceptually linked to the Theory of Planned Behavior [3]. The CAB model proposes that behavioral change unfolds through a sequential pathway: individuals’ cognition (what they know) influences their attitude (how they evaluate), which in turn shapes their behavior (what they do). Unlike economically deterministic models, CAB emphasizes the mediating roles of perception, emotional judgment, and social cues—factors that are especially influential in rural contexts characterized by uncertainty, peer imitation, and relational decision-making [4].

Nonetheless, the empirical application of the CAB model in agricultural contexts—particularly within large-scale, policy-driven interventions—remains underdeveloped. Existing studies often emphasize knowledge dissemination as the principal means of promoting adoption, while giving insufficient attention to how farmers form subjective attitudes or how institutional credibility influences behavioral outcomes. In practice, even technically robust initiatives like the “Lishu Model” may falter at the final stage of implementation—the behavioral “last mile”—where adoption depends not only on information access but also on internal belief formation and context-specific judgment.

This study applies the CAB framework to examine the behavioral mechanism underlying farmers’ adoption of conservation tillage in Lishu County, Jilin Province—the institutional birthplace of the “Lishu Model”. By integrating first-hand household survey data with contextual policy analysis, it explores how cognition and attitude jointly condition behavioral outcomes under a policy-driven intervention. While the study does not pursue causal inference due to data and methodological constraints, it aims to uncover how behavioral heterogeneity mediates the transmission of top-down ecological policies. By distinguishing between informed non-adopters and attitudinally aligned adopters, the analysis sheds light on the motivational pathways through which state-led green interventions translate—or fail to translate—into actual behavioral change.

2. Literature Review and Analytical Framework

While institutional support and economic incentives are widely recognized in technology adoption research, they alone cannot fully explain farmers’ decisions. Increasingly, scholars have turned to behavioral models to capture how cognition and attitude influence adoption under uncertainty. This section reviews that shift and introduces the CAB framework as the analytical basis for this study.

2.1. From Economic to Behavioral Logic

Traditional research on agricultural technology adoption has predominantly relied on economic models, which frame farmers as rational actors responding to material incentives. These models assume that decision-making hinges on maximizing expected utility—calculating the trade-off between input costs, anticipated yields, and available subsidies. While such frameworks provide a clear structure for policy design, they often fail to account for the complexity and unpredictability of real-world behavior, particularly in domains like conservation agriculture in which the outcomes of new practices are uncertain and in which adoption remains uneven despite strong institutional support.

A growing body of empirical evidence suggests that financial incentives and technical training, though necessary, are rarely sufficient to induce sustained behavioral change. Farmers’ decisions are often shaped by prior experience, perceived risk, trust in institutional actors, and local social dynamics. These findings have prompted a shift in the literature toward more behaviorally grounded approaches, recognizing that adoption is often mediated by interpretation, rather than determined solely by economic incentives.

This shift has opened the way for structured psychological models to enter the analytical mainstream. Among them, the Cognition–Attitude–Behavior (CAB) framework provides a useful entry point for examining how farmers process, evaluate, and act upon new information. Rather than treating behavior as a direct extension of financial calculus, CAB emphasizes the mediating roles of knowledge and subjective evaluation. It is especially well suited to contexts in which policy-led interventions aim to promote unfamiliar or long-horizon practices and the farmers’ responsiveness hinges less on material gain than on how they internalize policy cues.

The relevance of the CAB model is further underscored by recent studies demonstrating that variation in technology adoption is often driven not by resource constraints but by how farmers interpret institutional messages and make sense of new technologies within their local cognitive and emotional frameworks. In this sense, CAB enables a closer examination of the “behavioral last mile”—the often-overlooked stage where policy intentions must be transformed into individual decisions under conditions of uncertainty and social embeddedness.

2.2. The CAB Model as an Analytical Framework

The Cognition–Attitude–Behavior (CAB) model offers a sequential perspective on behavioral decision-making, positing that individuals’ knowledge (cognition) shapes their evaluative judgments (attitude), which in turn drive concrete actions (behavior). Rather than treating behavior as a direct outcome of incentives or regulation, CAB emphasizes the internal psychological processes through which individuals interpret, filter, and act upon external stimuli.

In the agricultural context, cognition refers to how farmers understand the functionality, risks, and institutional framing of new technologies, such as conservation tillage. Attitude captures their evaluative stance—whether they perceive the practice as necessary, beneficial, or aligned with their own goals or community norms. Behavior is then expressed in the concrete decision to adopt or reject the practice.

Compared with closely related models such as the Theory of Planned Behavior (TPB), CAB places a stronger emphasis on informational salience and affective resonance, making it particularly relevant for understanding responses to top-down policy interventions. In such settings, behavioral outcomes often depend less on material incentives than on how farmers internally process policy signals and frame the legitimacy or value of proposed practices.

Empirical applications of CAB in rural China have begun to shed light on these dynamics. A recent study on straw-returning technology adoption in major grain-producing provinces found that subjective cognition—not just awareness, but farmers’ self-assessed clarity and confidence—exerts a significant influence on attitudinal orientation, particularly when technical support mechanisms are in place [5]. Similarly, in the domain of digital agriculture, cognition—both subjective and objective—has been shown to enhance farmers’ enthusiasm for technology adoption, especially when embedded within visible institutional incentives such as training programs and subsidies [6]. These findings highlight the importance of aligning informational clarity with attitudinal readiness to facilitate behavioral transitions.

Building on this body of research, the present study employs the CAB model not only as a theoretical lens but also as an analytical foundation for variable construction and empirical modeling. By disaggregating cognition and attitude as separate yet interacting factors, we aim to uncover how farmers in Lishu County interpret and respond to conservation tillage initiatives under policy-led conditions. This approach provides insight into the micro-foundations of green technology adoption and helps identify where policy interventions can most effectively operate—whether by enhancing technical clarity, building psychological alignment, or embedding credibility in institutional delivery. Recent meta-analytical evidence further supports the behavioral relevance of CAB-type frameworks in agricultural sustainability, highlighting cognition and attitude as key conduits through which farmers internalize and respond to external policy cues [7].

3. Research Design and Descriptive Results

To empirically examine the behavioral mechanism proposed by the CAB framework, this section outlines the research design, data sources, and variable definitions used in the analysis. Based on a household survey conducted in Lishu County—the origin of the “Lishu Model”—we present an overview of the sample structure and variable construction, followed by descriptive evidence on how cognition and attitude relate to conservation tillage adoption. All statistical analyses were conducted using Stata 18 (StataCorp LLC, College Station, TX, USA). These findings set the stage for the regression-based analysis in the next section.

3.1. Data Source and Sample Overview

This study draws on an original household survey conducted in Lishu County, Jilin Province, between May and July 2024. As the pilot site for China’s nationally promoted “Lishu Model” for conservation tillage, Lishu offers both policy salience and mature implementation conditions. Six villages were purposively selected to capture variation in program exposure and local farming structures.

The questionnaire was developed in accordance with the CAB framework, comprising modules on demographic characteristics, technical cognition, attitudinal orientation, and actual adoption behavior. Out of 112 distributed questionnaires, 98 were deemed valid for analysis. The sample includes traditional smallholders, part-time farmers, and cooperative-affiliated households, ensuring diversity in farming experience and access to information.

The descriptive statistics show substantial heterogeneity within the sample. On average, the respondents were 58 years old, had seven years of formal education, and lived in households with four members. Approximately 38% reported membership in local cooperatives. This variation offers a solid empirical basis for exploring behavioral differences through the CAB model (

Table 1. Variable definitions and descriptions.

Variable	Description
adopt_lishu	Adoption of conservation tillage (1 = Yes, 0 = No)
know_lishu	Cognition level (1 = unaware, 5 = very familiar)
attitude	Attitude toward promotion (1 = not necessary, 5 = necessary)

).

In addition to the behavioral indicators, the sample’s demographic and structural features are summarized in

Table 2. Sample characteristics (n = 98).

Variable	Mean	Std Dev	Min	Max
Age	58.06	10.05	33.0	76.0
Education (years)	7.13	1.46	6.0	9.0
Household Size	4.01	1.55	1.0	8.0
Cooperative Member (%)	0.38	0.49	0.0	1.0

.

3.2. CAB Framework and Variable Design

The analytical foundation of this study lies in the Cognition–Attitude–Behavior (CAB) framework (Figure 1), which conceptualizes behavioral adoption as a sequential process. Rather than viewing action as a direct function of incentives, the CAB model underscores how internalized knowledge and evaluative judgment jointly shape behavioral outcomes.

Specifically,

Cognition is operationalized through a five-point scale reflecting farmers’ self-reported familiarity with the Lishu Model.

Attitude captures the perceived necessity of promoting conservation tillage locally.

Behavior is measured as a binary outcome: whether the respondent has adopted the practice or not.

This structure allows us to disentangle the relative contributions of awareness and evaluation in shaping behavioral response and to identify potential psychological bottlenecks in the policy transmission process.

3.3. Descriptive Results and Behavioral Pathways

Among the 98 valid respondents, 54% reported having adopted conservation tillage techniques associated with the Lishu Model. However, adoption rates exhibited clear variation along cognitive and attitudinal lines.

As shown in Figure 2, adoption rates rise significantly with more favorable attitudes: all farmers who rated the practice as “very necessary” (level 5) reported adoption, while those who viewed it as less necessary exhibited markedly lower uptake. This monotonic trend underscores the importance of attitudinal commitment as a behavioral trigger.

Cognitive variation was more concentrated, with most responses falling in the mid-to-high range (levels 3 and 4). Nevertheless, adoption was nearly universal among those reporting higher familiarity. Although limited data from lower cognition levels restrict our ability to draw robust conclusions about threshold effects, the pattern still suggests that moderate familiarity may be sufficient to facilitate uptake under supportive conditions.

3.4. Interactional Mechanism: Heatmap Analysis

To investigate whether cognition and attitude interact in shaping behavior, a two-dimensional heatmap was constructed (Figure 3), mapping adoption rates across all possible combinations of the five-point cognition and attitude scales. Due to limited sample sizes at the lowest and highest cognition levels, the heatmap focuses on levels 3 and 4, where the majority of responses are concentrated.

In Figure 3, the Y-axis (attitude) increases from bottom to top, and the X-axis (cognition) from left to right. The upper-right quadrant (high cognition, high attitude) shows the highest adoption rates, while the bottom-left (low cognition, low attitude) shows almost none. Importantly, adoption remains low among those with high cognition but low attitude (upper-left quadrant). This pattern identifies a distinct subgroup of “informed non-adopters”: technically aware farmers who nonetheless refrain from action due to motivational misalignment or contextual skepticism. This cognitive–affective disjunction supports the CAB framework’s emphasis on the joint effect of knowledge and motivation in behavioral change. Similar dynamics have been observed in China’s black soil region, where adoption decisions hinge less on structural support than on subjective interpretation of policy salience and trust in institutional credibility [8].

In sum, this section provides structured evidence supporting the behavioral relevance of both cognition and attitude, while highlighting their potential interaction. These descriptive results lay the groundwork for the regression analysis in Section 4, where the relative explanatory power of each component is formally assessed.

4. Empirical Results and Mechanism Interpretation

Building on the descriptive patterns outlined in the previous section, this section presents the empirical testing of the Cognition–Attitude–Behavior (CAB) framework. While earlier observations indicated clear associations between cognition, attitude, and conservation tillage adoption, they also revealed behavioral heterogeneity—particularly among informed farmers who remain hesitant to act. To further examine these dynamics, we estimate a series of regression models that assess whether and how cognition and attitude influence adoption, while also evaluating the robustness of these effects across different specifications.

4.1. Regression Estimation and Core Findings

To test the behavioral pathway proposed by the CAB model, we employ a logistic regression using a binary dependent variable that indicates whether the respondent has adopted the Lishu Model (1 = yes, 0 = no). The primary explanatory variables include self-reported cognition and attitude; both measured on a five-point ordinal scale. Additional controls include three county-level contextual indicators: the agricultural industrial integration index, the number of local cooperatives, and the output value of agri-related services—used to approximate structural support for adoption.

After excluding cases with missing values, the final sample includes 37 valid observations. While the sample size is limited, the model nonetheless provides indicative insights into the relative effects of psychological and structural variables. The main results are presented in

Table 3. Logistic regression results.

Variable	Odds Ratio (OR)	Coefficient	p-Value	Interpretation
Know_lishu (Cognition)	1.632	0.89	0.373	Positive, not statistically significant
Attitude	2.718	2.34	0.019	Significant positive effect (p < 0.05)
Fusion_index	1.014	0.07	0.946	No significant effect
Coop_count	1.002	1.58	0.114	Slightly positive, not significant
Agri_service_gdp	0.986	−0.42	0.676	No significant effect

below.

As summarized in Table 3, only the attitude variable reaches conventional levels of significance. To better visualize this contrast, Figure 4 below plots the estimated coefficients with 95% confidence intervals.

As visualized in Figure 4, the coefficient for attitude is clearly significant, while cognition—despite being positively signed—fails to reach statistical significance. This contrast reinforces the idea that awareness alone is insufficient; only when farmers perceive the practice as necessary does it translate into behavioral adoption.

4.2. Mechanism Interpretation: Beyond Individual Knowledge

The results suggest that awareness alone does not guarantee behavioral response. While farmers with higher cognition levels are somewhat more likely to adopt conservation tillage, the effect is not robust. In many cases, what is reported as “familiarity” may reflect a superficial understanding of the practice, lacking operational clarity or long-term cost–benefit perception. Additionally, farmers may be embedded in relational environments—such as peer networks or kinship-based production circles—in which individual initiative is constrained by collective hesitation.

Attitude, in contrast, demonstrates a decisive and consistent influence. Farmers who perceive the Lishu Model as necessary for local promotion are significantly more likely to adopt it. This validates the sequential logic of the CAB framework: cognition must be interpreted through a motivational lens before it translates into action. The result also aligns with the Theory of Planned Behavior, in which intention—shaped primarily by attitudinal dispositions—serves as the proximate driver of behavior.

The null findings for structural variables may reflect a disconnect between institutional presence and behavioral salience. While cooperatives and agricultural services exist at the county level, they may not be meaningfully integrated into smallholders’ decision-making contexts. This gap reinforces the importance of embedding institutional efforts into micro-level behavioral environments.

4.3. Robustness Checks

To validate the robustness of our findings, we conduct two additional tests: a Probit model estimation and a check for multicollinearity.

4.3.1. Probit Model Estimation

To assess robustness, we re-estimate the core model using a Probit specification. The results—summarized in

Table 4. Probit regression results.

Variable	Coef.	Std. Err.	z	p-Value	95% Conf. Interval
Intercept	−1.196	0.603	−1.98	0.047	[−2.378, −0.014]
Attitude	0.664	0.193	3.45	0.001	[0.286, 1.042]

—remain consistent with the Logit estimation, reinforcing the strength of the attitudinal effect.

As shown in Table 4, the Probit model confirms the robustness of the attitude effect. Figure 5 further compares the magnitude and precision of attitude estimates across both model types, highlighting their convergence.

In Figure 5, the coefficients for attitude across both Logit and Probit models are nearly identical in magnitude and significance, providing further evidence for the robustness of this psychological factor in influencing adoption decisions.

4.3.2. Multicollinearity Considerations

Although the sample size restricts the computation of variance inflation factors (VIFs), the conceptual independence of cognition and attitude, combined with their distinct coefficient trajectories across specifications, suggests that multicollinearity is unlikely to pose a major concern.

4.4. Policy Interpretation: Behavioral Alignment over Information Provision

While the study does not claim causal inference, the regression findings imply that boosting awareness may not be sufficient. Instead, behavioral response appears to hinge on whether farmers internalize the practice as necessary, feasible, and socially validated.

From this perspective, behavioral heterogeneity in adoption may reflect variations in emotional engagement, perceived legitimacy, and contextual resonance—rather than differences in mere knowledge levels. Therefore, efforts to promote sustainable practices like conservation tillage should prioritize not only knowledge dissemination but also mechanisms that foster attitudinal alignment.

In summary, this section provides empirical support for the CAB framework’s explanatory relevance. It also sheds light on the limits of top-down promotion strategies, highlighting the need for psychologically attuned and socially embedded interventions in advancing green agricultural practices.

5. Conclusions and Policy Recommendations

This study investigated the behavioral mechanisms influencing farmers’ adoption of conservation tillage under the “Lishu Model,” using the Cognition–Attitude–Behavior (CAB) framework as a conceptual lens. Drawing on original household survey data from Lishu County and combining descriptive insights with regression-based analysis, the findings reveal how cognitive familiarity and attitudinal disposition jointly shape adoption decisions in the context of state-led ecological interventions.

5.1. Key Findings

Three central insights emerge from the empirical analysis.

First, cognition alone is insufficient to explain adoption behavior. While descriptive statistics suggested a positive correlation between familiarity and uptake, this effect becomes statistically insignificant once attitudinal and contextual factors are considered. This indicates the presence of “informed non-adopters”—farmers who possess knowledge but remain reluctant to act. Such behavioral inertia echoes concern about motivational gaps among technically literate actors, as also raised in the prior literature and in reviewer feedback.

Second, attitude proves to be the most consistent predictor of adoption. Farmers who perceive conservation tillage as necessary for local practice are significantly more likely to implement it, even after controlling for structural conditions. This finding reinforces the CAB model’s emphasis on motivational alignment and offers empirical support for theories that prioritize intention over information as the proximate driver of behavioral change.

Third, structural variables such as cooperative presence and service infrastructure show no significant influence on adoption decisions. This suggests that institutional supply—though crucial in administrative design—does not automatically translate into behavioral engagement. Rather, as highlighted in previous research and reviewer suggestions, formal structures must be meaningfully embedded in farmers’ everyday decision contexts to be effective.

5.2. Policy Implications

Taken together, the results point to the need for a more behaviorally attuned approach to green technology promotion.

Reframe from knowledge transmission to motivational engagement: Behavioral change is unlikely to result from technical awareness alone. Policies should focus on how practices are perceived—whether they are seen as legitimate, feasible, and valuable in local contexts.

Strengthen peer-based demonstration mechanisms: Adoption is often shaped by social reference points. Village pilots, model households, and peer-to-peer testimonials can help reduce uncertainty and build emotional alignment. Extension personnel should be equipped not only with technical skills but also with communicative and relational capacity.

Embed institutional credibility in practice: Platforms like cooperatives and service stations must move beyond symbolic presence and become operationally relevant to farmers’ choices. Trust-building and accessibility should be integral to institutional design.

Diagnose attitudinal barriers early: Prior to program rollout, behavioral diagnostics—such as assessments of perceived relevance, trust, or risk—can help identify soft constraints and guide targeted interventions.

5.3. Limitations and Future Research

This study has several limitations. The regression analysis is based on a subsample of 37 complete observations, limiting statistical generalizability. Moreover, the indicators for cognition and attitude rely on single-item self-reports, which may not fully capture the complexity of these constructs.

Additionally, the CAB framework, while powerful, does not directly account for economic cost–benefit considerations—a limitation noted by reviewers. Future studies could explore hybrid models that integrate behavioral logic with economic reasoning, such as perceived profitability or risk aversion. The emerging literature in behavioral agricultural economics has called for such integrative approaches, emphasizing that motivational alignment and economic rationality often interact to shape real-world technology adoption decisions [9].

To deepen causal inference, experimental or longitudinal research is needed. Field trials incorporating attitudinal interventions (e.g., framing effects, trust signals, peer education) could test how shifts in cognition and attitude evolve and whether they yield lasting behavioral change.

5.4. Theoretical Contribution

Conceptually, this study extends the CAB framework to the domain of institutionalized agricultural modernization. It shows that, even in settings with strong policy and technical backing, adoption remains contingent on internalized evaluations. By highlighting the mediating role of attitude and the limits of structural provision, the analysis offers a more behaviorally grounded understanding of technology uptake in rural sustainability transitions.