Factors affecting Iranian farmers’ intention to adopt digital agriculture: an integrated TPB-TAM approach

Abstract

Digital agricultural technologies, with their proven potential to enhance productivity, resilience, and resource efficiency, play a key role in achieving food security and sustainable development. However, the adoption of these technologies in developing countries like Iran faces a significant limitation. Focusing on the existing knowledge gap. This study investigates the individual and social determinants of digital technology adoption among Iranian farmers by integrating the Theory of Planned Behavior and the Technology Acceptance Model. We apply this framework to targeted research with farmers of Kermanshah Province in western Iran, using survey data (sample n = 384) analyzed through structural equation modeling. Based on the results, the integrated model demonstrated strong explanatory power. Attitude and perceived behavioral control emerged as the strongest predictors of adoption intention, with attitude itself being largely shaped by perceived usefulness and ease of use. Contrary to expectations, subjective norm exerted no significant direct effect on intention, nor was it influenced by perceived usefulness. These findings reveal that the adoption process in this context is driven primarily by individual assessments of utility and feasibility rather than social pressures. This suggests a critical need to adapt classic adoption frameworks when applied to settings with weaker social capital or less cohesive institutional structures. Consequently, extension strategies should pivot from normative campaigns to approaches that emphasize personalized financial incentives, risk mitigation guarantees, and user-centric technology design.

Introduction

Food security has become a critical global concern due to rising hunger, imbalanced diets, and environmental degradation^1–3. It is a fundamental prerequisite for national security and economic development^4–6. Historically, agriculture focused on increasing production and income; however, overuse of inputs, water depletion, soil degradation, and rural decline have shifted its role. In the Fourth Industrial Revolution, agriculture faces two opposing trajectories: escalating threats from climate change, water scarcity, and population growth which have doubled the number of food-insecure people^3–9 and the emergence of digital agriculture as a transformative strategy to enhance productivity, optimize inputs, and strengthen resilience⁷. Digital agriculture technologies broadly defined as data-driven tools for precision resource management, farm automation, and information connectivity hold proven potential to enhance productivity and sustainability^7,8.

Despite this potential, a significant digital divide persists, particularly in developing countries⁹. This challenge is acute in western Iran, especially Kermanshah Province a major agricultural hub vital for regional food security and livelihoods. The area suffers from severe water stress and soil erosion¹⁰. Furthermore, agriculture in this region is closely intertwined with the Zagros mountain forests. Unsustainable farming practices, such as dryland cultivation on steep slopes and conversion of forest margins to cropland, have historically contributed to forest degradation. Conversely, forest loss exacerbates soil erosion, reduces water retention, and disrupts local climate regulation, creating a negative feedback loop that undermines long-term agricultural sustainability. Recognizing this interdependence is essential for understanding how agricultural interventions, including digital technologies, may have broader implications for forest conservation and landscape-level resilience. And although it holds above average shares in agricultural employment and value added, it has failed to achieve sustainable food security. A considerable portion of the population remains food insecure, yet the adoption of digital technologies among farmers is low and uneven^9,11.

This conflict between the urgent need and limited adoption presents a critical research and policy issue: What factors determine the willingness of farmers in the western region of Iran and Kermanshah province to adopt digital technologies? Answering this question is essential for designing effective and context-specific strategies to accelerate the digital transformation of agriculture in this area.

Studies show that the success of this transformation does not solely depend on access to technology, but also on a deep understanding of the psychological, social, and institutional factors influencing farmers’ decisions^12,13. In this context, the specific socio-economic fabric of the West of Iran, including strong social and tribal structures, tangible infrastructural limitations, and a lower level of institutional investment, necessitates looking beyond a single theoretical framework to analyze the adoption process.

On the one hand, the Technology Acceptance Model (TAM) focuses on individual perceptions such as usefulness and ease of use and well explains the cognitive mechanisms shaping attitudes. However, this model alone is not able to predict the influence of collective normative forces or the objective environmental constraints that are prominent in rural communities in Iran. On the other hand, the Theory of Planned Behavior (TPB), by introducing the concepts of subjective norms (the influence of social expectations) and perceived behavioral control (an individual’s perception of barriers and facilitators), offers a more comprehensive framework, but it may not be precise enough in measuring the nuances of technology-specific perceptions.

In general, although the TAM and TPB models have been widely used in technology adoption studies. Integrating these two models in the context of the study area seems not only useful but also necessary. In Kermanshah province, a farmer’s decision to adopt digital technology is likely not just the result of an individual calculation of profit and ease, but a complex mix of these calculations, influence from local leaders, peers, and the kinship network (subjective norm), as well as a realistic assessment of their ability to overcome barriers such as cost, technical knowledge, or access to support (perceived behavioral control). Therefore, an integrated framework allows for the simultaneous and interactive measurement of both the techno-cognitive and socio-structural aspects influencing farmers’ behavioral intentions.

Given that the combined application of TAM and TPB models in the agricultural context of Iran, especially in western regions with socio-tribal structures and weak social capital, has been less studied. The main gap of this research is to test the claim that in environments with inefficient institutions and low social trust, does the force of subjective norms (a key element of TPB) have an effect on the technology adoption process and is it replaced by the calculation of individual utility and perceived behavioral control?

By combining the two aforementioned models, this study seeks to fill this conceptual and contextual gap and answer the following questions: What individual and social factors determine the willingness of farmers in Kermanshah province to adopt digital agricultural technologies and what is the relative contribution of each variable in the adoption process?

The ultimate goal, beyond the scientific explanation of this process, is to provide context-specific and practical policy recommendations that can, by considering the region’s unique social, economic, and environmental characteristics, bring the wave of digital transformation to the plains and farms of this vital part of the country and play an effective role in strengthening national food security and sustainable agricultural development.

Theoretical framework and research hypotheses

The integration of the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB) in this study derives from a deliberate, context-driven theoretical selection, grounded in the complementary explanatory power each framework offers for examining technology adoption in complex agricultural settings. TAM provides a precise analytical lens on the cognitive determinants of acceptance namely perceived usefulness and perceived ease of use which are essential for understanding how farmers evaluate the instrumental value and usability of digital innovations. However, TAM alone does not account for the social and contextual constraints that frequently shape decision-making in rural communities. TPB, by incorporating subjective norms and perceived behavioral control, captures both the influence of social expectations and the farmer’s appraisal of internal and external barriers. Given the distinctive socio-institutional fabric of western Iran, characterized by dense kinship networks alongside low trust in formal institutions, neither model in isolation can adequately represent the multidimensional nature of adoption decisions. The integrated framework employed in this study enables the simultaneous examination of how techno-cognitive perceptions and socio-structural factors interact, and allows for the identification of which pathways predominate in a context where collective norms may be attenuated. This theoretical synthesis thus constitutes a methodological strategy explicitly aimed at enhancing both predictive accuracy and contextual validity.

Technology acceptance model (TAM)

The Technology Acceptance Model is one of the most effective approaches to the adoption of new technologies. According to Davis¹⁴, the Technology Acceptance Model seeks to explain and predict technology acceptance. This theory is an adaptation of the theory of reasoned action¹⁵, which determines attitudes toward behavioral intentions and the usefulness of technology use. The two core constructs of the Technology Acceptance Model are perceived usefulness and perceived ease of use. Perceived usefulness is defined as farmers’ view of the specific and practical application of the technology¹⁶, and perceived ease of use refers to the extent to which users expect the system to be effortless to use. These two beliefs (perceived usefulness and perceived ease of use) influence the user’s attitude toward using the system¹⁷. According to Davis’s¹⁴ original conceptualization of the Technology Acceptance Model, perceived usefulness and perceived ease of use do not directly determine behavioral intention in all contexts; rather, their influence is often channeled through attitude. Attitude represents the individual’s overall evaluative judgment positive or negative toward performing the behavior, which in turn shapes behavioral intention¹⁸.

This mediating role of attitude has received substantial empirical support in technology adoption research. For instance, Scherer et al.¹⁹, in their meta-analysis of TAM studies, concluded that attitude consistently mediates the relationships between technology perceptions and usage intentions. Similarly, Li et al.²⁰ found that among farmers considering smart agricultural technologies, the effects of perceived usefulness and ease of use on adoption intentions were partially transmitted through their attitudes toward the technology (Fig. 1). A study using the extended theory of planned behavior by Li et al.²⁰ showed that in the behavioral belief dimension, farmers valued the positive impact of perceived usefulness, even if the technology risk itself had a reducing effect on adoption intentions. In this regard, the results of Pishnyak and Khalina²¹ also showed that ease of use and usefulness had a positive impact on the willingness to use digital technologies. Furthermore, the results of a 2016 study showed that the growth in technology adoption is linked to economic gains in agriculture. Silva et al.²² conducted a study titled “Adoption of digital agricultural technologies in sugarcane production in sao paulo.” Their research results showed that companies that have adopted and used these technologies have demonstrated benefits such as improved management, higher productivity, lower costs, minimizing environmental impacts, and improved sugarcane quality. Lowenberg-Deboer and Erickson²³ also concluded in their research on precision agriculture adoption that initial economic returns affect the adoption of precision agriculture. Furthermore, research in southwestern Iran has shown that economic incentives alone may fail to convince farmers to adopt new technologies like drip irrigation, pointing to deeper behavioral and perceptual hurdles²⁴. Based on the results of the research Shambayati and Qahremani²⁵, three factors, data and information security, cost, and financial resources influence the adoption of digital technology in the health value chain. Hosseininiya et al.²⁶ showed that perceived usefulness, ease of use, trust, social influence and perceived behavioral control affect farmer’s willingness to adopt digital technologies.

Fig. 1.

Conceptual framework of the research.

Theory of planned behavior (TPB)

One of the most prominent theoretical frameworks in the field of understanding human behavior is the Theory of Planned Behavior, which was proposed by Ajzen and Fishbein²⁷. According to this theory, the most direct predictor of behavior is the intention (for example, a motive or plan) to engage in the behavior. Intention is an individual’s perception of engaging in a specific behavior; in other words, an individual’s intention is a mental and potential state that links the person to their action²⁵. In another definition, intention refers to the effort to perform a specific behavior as opposed to the actual behavior²⁸. It can be said that intention is a motivational factor that influences behavior and indicates how much effort individuals will make to perform a behavior²⁹. According to Davis¹⁴, the intention to perform a behavior is an important variable that can influence behavior, and it is necessary to determine an individual’s intention regarding a specific behavior before examining the behavior itself. In other words, based on an individual’s intention, their behavior can be predicted. Given that the policy of the Iranian government is to expand the use of digital agricultural technologies among farmers, the use of these technologies in Iran has not developed to the expected extent. Therefore, it is not possible to examine farmers’ behavior regarding the use of digital agricultural technologies. However, by examining farmers’ propensity for using digital agricultural technologies, their future behavior regarding the use or non-use of these technologies can be predicted. Based on this, the propensity variable is considered the dependent variable of the research (Fig. 1).

Intention, in turn, is predicted by three main factors: attitude, subjective norms, and perceived behavioral control³⁰. In the theory of planned behavior, attitude toward the behavior is a favorable or unfavorable evaluation of the behavior. For example, the portrayal of digital technology use as a desirable action reflects a positive attitude toward the adoption of digital agriculture. The influence of attitude on the intention to adopt has been confirmed based on the findings of studies by Li et al.²⁰, Bamberg et al.³¹, and Savari³².

Subjective norms assess whether individuals perceive social support from others in their lives to engage in a particular behavior. If people perceive that those around them consider using digital technology a worthwhile activity, they will consequently feel they have social support for using digital technologies, which will make them more likely to adopt the behavior. The impact of subjective norms on adoption was examined in the studies of Dissanayake et al.¹⁷(the impact of subjective norms on farmers’ perception of technology); Savari and Noushad³³(the impact of subjective norms on the intention to adopt water-saving technologies by farmers); and Savari³²(the positive impact of subjective norms on the intention to adopt environmental conservation behaviors by farmers), has been confirmed. However, the results of Li et al.²⁰ showed that peer influence had little effect on farmers’ willingness to adopt smart agriculture technology.

Perceived behavioral control is another variable in the theory of planned behavior that influences intention. Perceived behavioral control indicates the extent to which people think a behavior is something they can easily perform³⁴. For example, if someone cannot use a technology due to a lack of money, then they do not have real control over the behavior, and this lack of control directly affects their intention to use the technology. The impact of perceived behavioral control on intention to adopt has been confirmed in studies by Li et al.²⁰, Savari³², Klockne³⁵, Li et al.³⁶, Gao et al.³⁷, Tan et al.³⁸ and Mohammadi et al.³⁹ has also been confirmed. Therefore, based on the theory of planned behavior, if people have a positive attitude toward digital technology, if they perceive that other important people in their lives consider it a good thing to do, and if they believe it is an important matter that they can perform easily, Consequently, they will seek to adopt the technology, and in this process, their intentions should sequentially translate into actions of adopting agricultural digital technology³⁰.

Based on the theoretical foundations, the hypotheses of the study are as follows:

H1: Perceived usefulness has a positive and significant effect on attitude.

H2: Perceived ease of use has a positive and significant effect on attitude.

H3: Perceived usefulness has a positive and significant effect on subjective norm.

H4: Perceived ease of use has a positive and significant effect on subjective norm.

H5: Attitude has a positive and significant effect on intention to adopt;

H6: Subjective norm has a positive and significant effect on intention to adopt;

H7: Perceived behavioral control has a positive and significant effect on intention to adopt.

H8: Attitude mediates the relationship between perceived usefulness and the intention to adopt digital agricultural technologies.

H9: Attitude mediates the relationship between perceived ease of use and the intention to adopt digital agricultural technologies.

Methodology

Study population and sampling methodology

The study population comprised all farmers in Kermanshah Province (N ≈ 164000). The sample size was determined as 384 using Krejcie and Morgan’s⁴⁰ table for a finite population. A multi-stage stratified random sampling technique was employed to ensure representativeness:

Stage 1. (stratification by region): The province was divided into five agricultural zones (north, south, east, west, and central). One county was randomly selected from each zone, resulting in five counties.

Stage 2. (selection of districts and villages): From each selected county, one district was randomly chosen. Subsequently, one village was randomly selected from each chosen district, yielding five villages in total. Village selection considered the primary cultivated crops in the district to capture agricultural diversity.

Stage 3. (selection of farmers): From each final village, the required number of farmer participants was selected randomly from official agricultural household lists, proportionate to the village’s farmer population. This process ensured every farmer in the sampling frame had an equal chance of selection. Ultimately, 371 completed questionnaires were retained for analysis.

Study area

Kermanshah province, located in western Iran, is the seventeenth-largest province in terms of area (Fig. 2). Its fertile soil and suitable climate have transformed it into one of Iran’s agricultural hubs, producing strategic agricultural products such as wheat, barley, sugar beets, and chickpeas, making it one of the top-producing provinces in the country. Moreover, agriculture is the primary occupation for most rural residents in this province. However, despite these agricultural advantages, Kermanshah province, does not enjoy a favorable position in terms of food security. Studies have indicated that Kermanshah is one of the top three provinces in the country facing food insecurity^41,42. Importantly, agriculture in Kermanshah province is not an isolated activity but is closely intertwined with the region’s natural ecosystems, particularly the zagros mountain forests. The zagros woodlands, which cover substantial parts of western Iran, provide essential ecosystem services including water regulation, soil conservation, and climate moderation, all of which directly support agricultural productivity in adjacent farmlands^41,42. Conversely, traditional farming practices, such as dryland farming on sloping lands, overgrazing, and the conversion of forest margins to cropland, have contributed to forest degradation and soil erosion in the region^43,44. This mutual dependence means that agricultural policies – including those promoting digital technologies can have significant indirect consequences for forest health and biodiversity conservation. Recognizing this interconnection is essential for designing sustainable development strategies that jointly address food security and environmental protection.

Fig. 2.

Location of the study area.

To address food insecurity in the province, researchers have suggested the adoption of DAT^43,44. Given that Kermanshah is one of the influential provinces in the country in the field of agriculture, the adoption of DAT by farmers in this province can significantly impact provincial and national food security. Therefore, Kermanshah province in western Iran was selected as the study area.

Defining digital agriculture technology

For the purposes of this study, “digital agriculture technology” (DAT) is operationally defined as a suite of tools, systems, and applications that collect, process, and analyze digital data to inform and optimize agricultural decision-making across the production cycle. Drawing on the conceptual frameworks of the Food and Agriculture Organization^7,8, DAT encompasses technologies that enable:

1. Precision resource management: Technologies such as variable-rate irrigation systems, soil sensors, and drone-based crop monitoring that allow for site-specific application of water, fertilizers, and pesticides.

2. Data-driven farm management: Platforms including farm management information systems (FMIS), mobile agricultural advisory services, and cloud-based record-keeping tools that support planning, monitoring, and traceability.

3. Automation and smart equipment: Innovations such as automated irrigation controllers, smart greenhouses, and GPS-guided machinery that reduce manual labor and enhance operational efficiency.

4. Market and information connectivity: Digital platforms that provide real-time market prices, weather forecasts, pest alerts, and linkages to input suppliers or buyers.

This operational definition guided the selection and development of measurement items in this study. Farmers were asked to consider these categories of technologies when responding to questions about perceived usefulness, ease of use, attitude, and adoption intentions. The definition ensures conceptual consistency between the theoretical construct under investigation and the empirical indicators used to measure it.

Research instruments

A two-part questionnaire was employed for data collection. The first section gathered information on the demographic characteristics of farmers, including gender, age, education level, farming experience, household size, religious knowledge, and participation in training courses related to digital agricultural technologies. The second part of the questionnaire consisted of 30 items to determine the components of the model’s components, with perceived ease of use measured by 5 items, perceived usefulness by 5 items, attitude by 5 items, subjective norm by 5 items, perceived behavioral control by 3 items, and intention to adopt by 7 items. Respondents were asked to indicate their level of agreement or disagreement with each item using a Likert scale from 1 (very low) to 5 (very high). The measurement items for perceived usefulness and perceived ease of use were deliberately crafted to reflect the operational definition of DAT provided. For example, items addressing water savings (V1), cost reduction (V2), reduced input usage (V3), and yield improvement (V4) correspond directly to the precision resource management and data-driven management dimensions of DAT. Similarly, items related to accessibility of information channels (S3) and technical support (S5) align with the connectivity and support infrastructure essential for DAT adoption. To ensure the appropriateness of the instrument within the developed Resistance to Innovation model, previous studies were used as references (Table 4). Before data collection, all respondents were assured that their information would remain completely confidential. Additionally, to maintain confidentiality, the questionnaires were anonymous.

Table 4.

Summary of goodness of fit indices for the measurement model according to Table 5, the square root of the AVE for each construct (diagonal values) is larger than the correlations between that construct and any other construct (off‑diagonal values). The diagonal values range from 0.692 to 0.884, and the off‑diagonal correlations range from 0.415 to 0.792. This confirms the discriminant validity of the constructs in the proposed model.

Fit index	SRMR	D-G1	D-G2	NFI	RMS-theta
Suggested value	< 0.1	> 0.05	> 0.05	> 0.90	≤ 0.12
Estimated value	0.07	0.364	0.439	0.96	0.09

Statement

All interviewees were informed about data protection issues by the enumerators and gave their consent orally at the beginning of each interview. Informed consent was obtained from all individual participants included in the study. All materials and methods are performed in accordance with the instructions and regulations and this research has been approved by a committee at college of agriculture & natural resources, razi university, Kermanshah, Iran. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Validity and reliability of the research instrument

The survey instrument was developed by adapting established scales from prior technology acceptance and planned behavior literature^14,20,28, to the context of digital agriculture. Content and face validity were ensured through a review by a panel of nine experts in agricultural extension, rural sociology, and environmental management. Their feedback led to refinements in wording, sequencing, and relevance. Construct validity and reliability were then assessed statistically. As shown in Table 1, all factor loadings were significant (p < 0.01). Another criterion for assessing reliability, which has advantages over the traditional Cronbach’s alpha method, is composite reliability (CR)⁴². To evaluate the validity of the research, in addition to expert opinions, convergent validity was used. Fornell and Larcker⁴⁵ proposed using the average variance extracted (AVE) as a criterion to assess convergent validity. As shown in Table 4, the values for both Cronbach’s alpha and CR indices are at a desirable and acceptable level (above 0.7), indicating the sufficient reliability of the items. The average variance extracted (AVE) for the latent variables are also above 0.5⁴⁶, indicating that all latent variables in the proposed research model have good reliability and validity. For discriminant validity, the Fornell-Larcker criterion and the heterotrait-monotrait (HTMT) ratio were applied⁴⁷ (Tables 5 and 6).

Table 1.

Results of confirmatory factor analysis for the measurement model.

Perceived ease of use (PEU)48		AVE = 0.674	CR = 0.876	Cronbach’s alpha = 0.876		ƛ	t
Using digital agricultural technologies is simple for farmers (S1)						0.878	10.269
It is easy for farmers to learn how to use digital agricultural technologies (S2)						0.743	8.449
nformation channels on digital agriculture are easily accessible (S3)						0.918	17.253
The interpretation of outputs related to digital agricultural technologies is clear and understandable (S4)						0.712	8.073
When facing a problem, technical support and advice to resolve it are easily available while working with digital agricultural technologies (S5)						0.833	16.435

Perceived usefulness(PE) 48		AVE = 0.536	CR = 0.782	Cronbach’s alpha = 0.731		ƛ	t
Using digital agricultural technologies saves water (V1)						0.728	2.308
Using digital agricultural technologies reduces production costs (V2)						0.859	2.313
The use of digital agricultural technologies leads to the use of fewer inputs (pesticides and chemical fertilizers) (V3)						0.639	1.387
The use of digital agricultural technologies increases crop yield (V4)						0.692	1.993
The use of digital agricultural technologies improves production and marketing methods (V5)						0.725	1.973

Attitude (Att)28		AVE = 0.542	CR = 0.784	Cronbach’s alpha = 0.726		ƛ	t
I think that the advantages of digital agricultural technologies far outweigh their disadvantages (N1)						0.849	14.350
I think that it is necessary for me to adopt digital agricultural technologies (N2.						0.777	25.669
I think that digital agricultural technology is a technology for the development and improvement of the future of agriculture (N3)						0.547	14.452
I think that using digital agricultural technologies increases my income by increasing crop production (N4)						0.669	34.686
I think that using digital agricultural technologies leads to the lower consumption and better management of inputs, and through this, I can save on costs (N5)						0.797	18.418

Subjective norm (SN)10		AVE = 0.657	CR = 0.861	Cronbach’s alpha = 0.784		ƛ	t
I should use digital agricultural technologies because my relatives, friends, and neighbors want me to use these technologies (H1)						0.745	0.894
I should use digital agricultural technologies because experts from the Agricultural Extension Organization want me to use these technologies (H2)						0.847	11.915
I should use digital agricultural technologies because society expects me, as a farmer, to use digital agricultural technologies (H3)						0.903	10.010
If I use digital agricultural technologies, my peers will approve of me (H4)						0.907	7.949
Because friends, relatives, and other farmers use some digital agricultural technologies, I should use them too (H5)						0.582	1.040

Perceived behavioral control (PBC)20		AVE = 0.781	CR = 0.860	Cronbach’s alpha = 0.825		ƛ	t
I have the conditions for adopting digital agricultural technology (cost, time, knowledge, etc.) (K1)						0.828	38.806
I can easily access the necessary knowledge and information about digital agricultural technology (K2)						0.891	21.265
I can easily adopt digital agricultural technologies (K3)						0.927	4.112

Intention (Int)10		AVE = 0.617	CR = 0.892	Cronbach’s alpha = 0.782		ƛ	t
I intend to recommend the use of digital agricultural technologies to other farmers (T1)						0.875	13.079
I would like to learn how to use digital agricultural technologies (T2)						0.751	23.518
I intend to participate in training classes and briefings related to digital agricultural technologies (T3)						0.894	13.731
I will use digital agricultural technologies in the future (T4)						0.813	34.218
I am willing to use digital agricultural technologies to increase productivity and reduce food insecurity (T5)						0.691	15.735
I would like to use digital agricultural technologies in the future (T6)						0/714	3.500
I plan to use digital agricultural technologies in the next farming season (T7)						0.720	3.133

Table 5.

Discriminates validity (Fornell-Larcker criterion).

Constructs	1	2	3	4	5	6
1. Int	0.786 a
2. PEU	0.460**	0.821 a
3. PU	0.467**	0.737**	0.797 a
4. ATT	0.494**	0.792**	0.657**	0.811 a
5. SN	0.586**	0.415**	0.429**	0.648**	0.884 a
6. PBC	0.663**	0.639**	0.583**	0.579**	0.581**	0.692a

**Correlation is significant at the < 0.01 level. At the square roots of AVE estimate.

Table 6.

HTMT result.

Constructs	1	2	3	4	5	6	7	8	9
1. Int
2. PEU	0.84
3. PU	0.73	0.36
4. ATT	0.82	0.74	0.66
5. SN	0.49	0.81	0.74	0.35
6. PBC	0.78	0.75	0.46	0.53	0.78

Following established guidelines for reflective measurement model assessment⁴⁷, indicator reliability was evaluated through standardized factor loadings. While the preferred threshold is 0.708, loadings between 0.40 and 0.708 were retained when: (a) composite reliability and average variance extracted for the respective construct already exceeded recommended thresholds (CR ≥ 0.70, AVE ≥ 0.50), and (b) removal of the indicator did not yield a substantial improvement in these metrics. This approach ensures both statistical reliability and content validity are preserved⁴⁷. All retained indicators with loadings below 0.708 were carefully evaluated and justified on these grounds. According to the results, the composite reliability (CR) and cronbach’s alpha values of the model components were above 0.6 and 0.7, respectively. Additionally, the average variance extracted (AVE) for all constructs in the proposed research model was above 0.5. These results indicate that all latent variables in the proposed research model have good reliability and validity (Table 1).

Data analysis

SPSS23 was used for descriptive analysis, including frequency, percentage, mean, and standard deviation.

SEM was used along with Smart-Pls4 software to test research hypotheses. SEM is a combination of structural and measurement models¹⁰¹, and it was preferred due to its advantages in providing a powerful multivariate analysis technique, similar to the family of multiple regressions. Another advantage of SEM is its consideration of measurement errors in the analysis⁴⁷.We conducted consistency analysis, including convergent validity (AVE), composite reliability (CR), and Cronbach’s alpha coefficient, using SmartPls4 software.

Results

Descriptive statistics

As listed in Table 2, the respondents’ average age was 43.56 years. The majority had elementary education, constituting 32.61%, while only 12.93% had college/university degrees. The mean household size and farming experience were four individuals and 34.28 years, respectively. The mean monthly household income was $140.32. About 89.48% of respondents stated that they had never participated in any training courses related to digital agriculture. Additionally, 83.53% of respondents considered their religious knowledge to be average or above average.

Table 2.

Demographic characteristics of farmers All model variables exhibited mean values above the theoretical midpoint (3), indicating an overall tendency toward higher-than-average responses (Table 3).

Variable	Category	Frequency(371)	Percent	Mode
Age	Lower than 30	75	20.21
	30–50	197	53.09	*
	More than 50	99	26.68
Education	Illiterate	50	13.47
	Elementary	121	32.61	*
	Secondary	85	22.91
	High school	67	18.05
	College education	48	12.93
Number of household (person)	Lower than 3	66	17.78
	3–4	179	48.24	*
	More than 4	126	33.96
Annual income (dollars)	Lower than 1000	61	16.44
	1000–2000	136	36.65	*
	2000–3000	93	25.06
	More than3000	81	21.83
Work experience	Lower than 20	111	29.91
	20–40	229	61.72	*
	More than 40	31	8.35
Presence in relevant training courses	yes	332	89.48	*
	NO	39	10.51
Knowing yourself religious	Very low	14	3.77
	Low	47	12.66
	Middle	199	53.63	*
	high	66	17.78
	Very high	45	12.12

Table 3.

Descriptive statistics of latent variables.

Variables	Mean (of 5)	Sd
Intention (Int)	3.24	0.452
Perceived ease of use (PEU)	3.77	0.654
Perceived usefulness (PU)	4.2	0.538
Attitude (Att)	3.83	0.632
Subjective norm (SN)	3.74	0.478
Perceived behavioral control (PBC)	3.08	0.874

Subsequently, the proposed conceptual model was evaluated in two stages: the measurement model assessment and the structural model assessment using the partial least squares (PLS) approach.

Measurement model: Confirmatory factor analysis (CFA) was conducted on various indicators to assess the validity and reliability of the research constructs, with the values presented in Table 4, indicating that the model demonstrates a good fit.

Table 6 shows that all HTMT values fall below the 0.85 threshold, providing additional support for the discriminate validity of the constructs.

Common method bias

Because all data were collected via self-report questionnaires at a single point in time, there is a potential risk of common method bias (CMB) that could artificially inflate or deflate observed relationships⁶³. To assess the severity of this threat, we employed two complementary diagnostic procedures.

First, Harman’s single-factor test was conducted using unroasted exploratory factor analysis with principal axis factoring. All 30 measurement items were entered, and the extraction was constrained to a single factor. The single factor accounted for 32.8% of the total variance, well below the conventional threshold of 50%, suggesting that CMB is unlikely to be a major concern⁶³.

Second, we applied the full co linearity assessment approach recommended by Kock⁶⁴ for PLS-SEM. A latent variable with a single randomly generated indicator was included as a common method factor predicting all endogenous constructs. Variance inflation factors (VIFs) were calculated for each substantive latent variable.

To assess multi co linearity among indicators, Variance inflation factor (VIF) values were computed for all reflective constructs. Following the guidelines of Hair et al. (2021), VIF values below 3.0 indicate no co linearity concerns (VIF < 3No co linearity concerns: VIF 3–5Potential co linearity issues, usually uncritical: VIF ≥ 5Critical co linearity issues likely present). As shown in Table 5, all outer VIF values ranged from 1.876 to 2.891, well below the conservative threshold of 3.0. These results confirm that indicator co linearity does not bias parameter estimates in our measurement model” (Table 7).

Table 7.

Full co linearity VIFs (common method bias assessment).

Construct	VIF
Perceived ease of use (PEU)	2.345
Perceived usefulness(PU)	2.567
Attitude (Att)	2.891
Subjective norm (SN)	1.987
Perceived behavioral control (PBC)	2.123
Intention (Int)	1.876

Based on these diagnostic tests, we conclude that common method bias does not constitute a threat to the internal validity of our findings.

After confirming the measurement models using CFA, the path analysis method (structural model assessment) was employed to test the hypotheses within the framework of the proposed conceptual model. The research path model, showing the standardized factor loadings, is presented in Fig. 3.

Fig. 3.

Path model with standardized factor loadings.

Hypothesis Testing: The results of the final effects of the variables on the intention to use Digital agricultural technologies (DAT) are presented in Table 7. The bootstrapping method was used to test the research hypotheses. According to the results, all research hypotheses were significant and confirmed except for the effect of the Perceived Usefulness on intention to use (Hypothesis 3) and the effect of the Subjective Norm on intention to use (Hypothesis 6) (Table 8).

Table 8.

Results of research structural models.

Hypothesis		T	Sig	Result
H1: PU ➞ ATT	0.278	2.22	0.011	Confirm
H2: PEU ➞ATT	0.647	2.38	0.000	Confirm
H3: PU ➞ SN	0.279	1.84	0.101	Reject
H4: PEU ➞ SN	0.199	2.30	0.004	Confirm
H5: ATT ➞ Int	0.981	5.16	0.000	Confirm
H6: SN ➞ Int	0.024	0.748	0.184	Reject
H7: PBC ➞ Int	0.391	3.21	0.000	Confirm
H8:PU → Att → Int	0.273	2.707	0.016	Confirm
H9:PEU → Att → Int	0.635	2.819	0.009	Confirm

Table 8 shows the path coefficients and t-values of the research hypotheses. Based on the t-values, two of the seven hypotheses under investigation were rejected.

Discussion

The foundational This study was conducted to identify the factors influencing the willingness of farmers in Kermanshah province, western Iran, to adopt digital agricultural technologies, by integrating the theory of planned behavior (TPB) and the technology acceptance model (TAM). The selection of a combined framework stems directly from the unique social, economic, and institutional characteristics of Kermanshah province. The results of this study indicate that the technology adoption process in this region is multifaceted and context-dependent. The Kermanshah farmer, in their decision-making, not only engages in an individual calculation of usefulness and ease of use (the key components of TAM) but simultaneously considers their perception of the ability to overcome practical and financial barriers namely, perceived behavioral control from the TPB. The key finding of the non-effect of subjective norms confirms and validates this integrated approach. This result indicates that despite the existence of dense social structures in rural communities, the low level of social capital and trust in institutions has neutralized the force of collective norms in the realm of new technologies. Therefore, adoption is more a function of personal evaluations (attitude) and perceived barriers (behavioral control) than of social pressure. This theoretical integration allowed for the identification of both effective direct and indirect paths and provided a scientific explanation for why one social path subjective norm was non-significant in this local context.”

The integrated TAM‑TPB model explained 86.4% of the variance in farmers’ intention to adopt digital agricultural technologies. While this value is higher than typical R² values reported in stand‑alone TPB studies, it is theoretically justifiable for three reasons. First, attitude fully mediates the effects of perceived usefulness and perceived ease of use on intention. This means that nearly all cognitive evaluations of the technology are channeled through a single attitudinal pathway, which concentrates explanatory power. Second, the intention construct was measured with high internal consistency, reducing unsystematic measurement error and allowing the structural predictors to account for a larger share of systematic variance. Third, the study sample was drawn from a relatively homogeneous farming population facing shared environmental and economic pressures (water scarcity, soil degradation, limited institutional trust). This contextual homogeneity reduces extraneous between‑respondent variance, thereby increasing the proportion of variance that can be explained by the modeled psychological constructs. Taken together, the high R² reflects the theoretical synergy of integrating TAM and TPB, and it underscores the pivotal role of attitude as a psychological gateway for technology adoption decisions in this context.

The results showed that among the examined paths, the attitude and perceived behavioral control variables had a positive and significant direct effect on the intention to adopt. The findings further revealed that attitude serves as a significant mediator in the relationships between both perceived usefulness and perceived ease of use with farmers’ intention to adopt digital agricultural technologies. These mediation effects (H8 and H9) were confirmed through bootstrapping analysis, with indirect effect coefficients of β = 0.273 (p < 0.05) for the PU→ Att → Int path and β = 0.635 (p < 0.01) for the PEU → Att → Int path (see Table 7).

This pattern of results aligns with logic of the Technology Acceptance Model (TAM), which posits that external variables influence behavioral intention primarily through their impact on attitude¹⁴. The findings demonstrate that farmers do not form adoption intentions based solely on cognitive evaluations of a technology’s usefulness or ease of use in isolation. Rather, these perceptions must first shape a favorable or unfavorable overall evaluation of the technology i.e., attitude which then translates into behavioral intention. This mediation pathway underscores the pivotal role of attitude as a cognitive-affective bridge between technology perceptions and behavioral outcomes.

The strong mediation effect observed for perceived ease of use (β = 0.635) is particularly noteworthy. In the context of Kermanshah’s farming community, where literacy levels and prior technological exposure may be limited, the perception that a technology is easy to learn and operate appears to be a powerful driver of positive attitude formation. This finding resonates with Scherer et al.¹⁹ meta-analytic conclusion that ease of use often exerts its influence on technology acceptance indirectly through attitude, especially among populations with lower technological self-efficacy.

For perceived usefulness, the mediation effect (β = 0.273), while smaller in magnitude, remains theoretically significant. This suggests that although usefulness directly shapes attitude (as confirmed in H1), part of its influence on adoption intention is channeled through the positive affective responses it generates. Farmers who recognize the tangible benefits of digital technologies such as water savings, cost reduction, or yield improvement develop more favorable attitudes, which in turn strengthen their adoption intentions. This finding is consistent with Li et al.²⁰, who reported similar mediation dynamics among Chinese farmers considering smart agricultural technologies.

Theoretically, these mediation results reinforce the conceptualization of attitude as a central mediating mechanism within integrated TAM-TPB frameworks. They demonstrate that the two models are not merely additive but interact synergistically: TAM’s technology-specific perceptions feed into TPB’s attitude construct, which then combines with other TPB components (subjective norms and perceived behavioral control) to shape intention. This integration captures the full psychological process by which external technology characteristics translate into internal motivational states and ultimately into behavioral commitment.

From a practical standpoint, these findings carry important implications for intervention design. Extension programs cannot assume that simply demonstrating a technology’s usefulness or ease of use will directly translate into adoption intentions. Rather, such demonstrations must be deliberately structured to foster positive attitudes for example, through hands-on experience, testimonies from respected peers, and opportunities for trial use that allow farmers to personally experience the technology’s benefits. Attitude change, as the mediation analysis confirms, is the critical intermediate outcome that ultimately drives adoption decisions.

The effect of perceived usefulness on subjective norm, as well as the effect of subjective norm on intention, was not significant. Consistent with Davis’s¹⁶ perspective, usability refers to the degree of ease of using the technology. The findings of this study, in line with the results of Li et al.²⁰ and Pishnyak and Khalina²¹, indicate that farmers’ perception of the ease of using and learning the technology positively influences their attitude. Similarly, perceived usefulness, which is the belief that the technology is applicable and beneficial¹⁸, also played a key role in shaping a positive attitude. This finding indicates that farmers’ rational judgment about the perceived benefits and the technology’s ease of implementation forms the cornerstone of developing a favorable attitude. This indicates the dominance of the instrumental rationality paradigm in the initial evaluation of innovation. This finding is consistent with studies by Li et al.²⁰, Pishnyak and Khalina²¹, Lowenberg-DeBoer and Erickson²³, Borghi et al.⁶⁵, and Silva et al.²². Given these results, extension programs should emphasize practical and tangible demonstrations of a technology’s efficiency and simplicity in pilot farms, going beyond mere information dissemination. The development of localized technologies with a very user-friendly interface and the conduct of practical, field-based training courses can enhance the positive perception of these two factors.

Another variable confirmed to influence the intention to adopt technology based on the results is attitude. As a general positive or negative evaluation of the behavior, attitude served as a key mediating construct in this model. According to Li et al.²⁰, farmers’ beliefs and inclinations can be influenced by their positive and negative attitudes and subjective beliefs. This finding confirms that technology characteristics (usefulness and ease of use), without shaping an overall positive evaluation (attitude), cannot be a direct driver of behavioral intention. Therefore, extension strategies should first focus on designing interventions that directly shape and improve attitudes. Emphasizing direct economic benefits (cost reduction, increased yield) and indirect benefits (reduced work effort, more precise management) through effective messaging can foster this positive attitude.

The influence of the two variables, perceived usefulness and ease of use, on the intention to adopt, mediated by attitude, indicates that technology features alone cannot directly drive behavioral intention without a positive overall evaluation. The impact of attitude on the willingness to adopt was confirmed, based on the findings of Li et al.²⁰, Bamberg et al.²⁹, and Savari³². Given this finding, it is suggested that extension programs should, in the first instance, focus on designing attitude-shaping interventions rather than merely promoting technical features (programs that directly influence farmers’ beliefs and overall evaluation of the technology and tangibly enhance their perception of usefulness and ease of use, rather than simply listing technical specifications). Such as establishing demonstration-trial farms, holding hands-on field workshops, conducting field visits, developing limited trial kits, providing financial simulations, and offering objective calculations).

Another variable that had a direct impact on the intention to adopt was perceived behavioral control. Perceived behavioral control refers to an individual’s perception of their ability to perform a behavior (e.g., due to the presence or absence of sufficient resources like knowledge or capital), and it represents the importance of objective and perceived barriers facing farmers. This result is consistent with the finding of the study by Li et al.²⁰. Therefore, it can be said that if farmers have a positive perception of the technology, its positive effects, and its ease of use, and if they also have the financial ability to purchase the technology, this combination of perceptions and attitudes can lead to an enhancement of farmers’ self-efficacy and increase their willingness to adopt and use the technology. To enhance the sense of control and self-efficacy, it is essential to design incentive packages and financial guarantees (such as targeted subsidies, long-term low-interest loans, and crop insurance related to new technologies). Additionally, strengthening support networks, after-sales services, and local maintenance can increase the sense of feasibility. Based on the findings, the following suggestions are made:

1. Designing personalized financial incentives: Move beyond uniform subsidies. Packages could include: (a) tiered purchase grants based on farm size and crop type, (b) vouchers for discounted digital subscription services from approved providers, and (c) tax rebates linked to verified reductions in water or chemical input usage achieved via technology adoption.

2. Implementing risk-mitigation guarantees: Establish a DAT adoption guarantee fund, backed by government or agricultural cooperatives, to cover yield losses demonstrably linked to technology failure in initial seasons. Parallels, promote vendor-backed buy-back guarantees for a trial period to reduce upfront investment risk.

3. Specifying low-complexity technology features: Technologies promoted should prioritize: (a) intuitive, icon-based user interfaces in local languages, (b) minimal required training for basic operation, (c) compatibility with widely available smart phones rather than proprietary hardware, (d) offline functionality for areas with poor connectivity, and (e) accessible local technical support networks.”

The non-significant effect of subjective norm on adoption intention, while seemingly counterintuitive in a collectivist social setting, can be theorized through the lens of social capital and institutional trust^66,67. For normative influence to be effective, it must be underpinned by trust in the institutions and peers promoting the norm. In the study region, evidence points to eroded institutional trust^68,69. When recommendations from extension agents or cooperative leaders are met with skepticism due to past experiences, the associated normative pressure loses its potency. Consequently, farmers may resort to a form of strategic individualism, where pressing livelihood risks and economic calculations outweigh ambiguous social expectations. This finding does not invalidate TPB but rather highlights a critical boundary condition: the subjective norm construct may be neutralized in environments where formal and informal institutions lack credibility, shifting the weight of behavioral intention entirely onto personal attitude and perceived behavioral control. To bridge this institutional gap, policies should be directed toward the genuine empowerment and identity-building of local grassroots and agricultural organizations. Transparency, accountability, and the fight against corruption within relevant institutions, as well as the effective involvement of leading and locally trusted farmers in the extension process, can gradually pave the way for the revival of trust and the re-emergence of positive social norms.

Overall, the research findings in Kermanshah province revealed the multidimensional nature of farmers’ decision-making. On one hand, the strong influence of attitude and perceived behavioral control (from the TPB) indicates that farmers, in addition to evaluating the usefulness of the technology (TAM), also realistically consider local practical and financial barriers. On the other hand, the non-significance of the subjective norm variable indicated that in the specific context of Kermanshah Province, with its characteristics such as high economic pressure and possibly weak social capital, social pressures do not play a central role. Therefore, the integrated model made it possible to examine both individual calculations and objective environmental constraints simultaneously, whereas a single model was unable to explain this pattern.

Research implications

Theoretical implications

This study provides three key theoretical contributions. First, the empirical finding that subjective norms do not significantly influence farmers’ adoption intention challenges the universal applicability of the theory of planned behavior (TPB) in contexts characterized by weak institutional trust and low social capital. In such settings exemplified by Kermanshah Province normative social pressures appear to be neutralized, and individual rational calculation takes precedence. This suggests a need for theoretical revision regarding the weight of the subjective norm construct in technology adoption models applied to similar socio-institutional environments.

Second, the strong direct effects of attitude and perceived behavioral control on intention, combined with the finding that attitude itself is primarily shaped by perceived usefulness and ease of use, confirm that adoption decisions in this context are driven by instrumental rationality. Attitude functions as a techno-cognitive construct rather than a socio-affective one, meaning that farmers evaluate technologies based on tangible benefits and operational feasibility rather than emotional or social considerations.

Third, the mediation analysis reveals that attitude fully transmits the effects of perceived usefulness and ease of use onto intention. This underscores the central mediating role of attitude in the integrated TAM-TPB framework. Furthermore, the finding that perceived usefulness does not significantly influence subjective norms indicates that individual cost-benefit logic does not easily translate into collective normative systems. This points toward strategic individualism in farmers’ decision-making under economic pressure and raises new theoretical questions about the nature of social capital formation in economically stressed rural contexts.

Practical implications

The empirical findings directly inform actionable policy recommendations. The non-significance of subjective norms implies that traditional extension approaches relying on group pressure, mass briefings, or indirect peer persuasion will be ineffective in this context. Instead, the strong influence of perceived behavioral control and attitude indicates that interventions must focus on enhancing farmers’ sense of personal capability and positive outcome evaluations.

Based on these findings, we propose a three‑pillar strategy:

Technology screening (informed by the strong effects of perceived usefulness and ease of use on attitude): Extension agencies should develop a context‑specific adoptability index that prioritizes technologies with demonstrable ease of use and tangible benefits. Technologies with low operational complexity and clear water‑saving or cost‑reduction benefits should be selected first.

Early‑adopter support (informed by the strong effect of perceived behavioral control on intention): Given that farmers’ perception of their ability to overcome financial and knowledge barriers strongly predicts adoption intention, policymakers should provide targeted financial safety nets–including low‑interest loans, risk guarantees, and subsidized technical support – to a select group of lead farmers. Creating visible success stories will enhance perceived behavioral control among the wider community.

Networked learning (informed by the mediating role of attitude): Since attitude fully transmits the effects of usefulness and ease of use onto intention, extension programs should shift from mass training to small‑group, trust‑based learning circles centered on early adopters. Hands‑on demonstrations and peer observation will directly shape positive attitudes more effectively than abstract information campaigns.

This paradigm shift from mass dissemination to smart facilitation is directly mandated by the finding that subjective norms do not drive adoption intention in this context. Importantly, these practical strategies also carry broader environmental implications, particularly for forest conservation in the zagros region, as explained below.

While our empirical model focused on individual adoption decisions, the findings carry indirect implications for landscape‑level sustainability, including forest conservation in the Zagros region. As noted in the study area description, agriculture and forests in Kermanshah are tightly coupled. Unsustainable farming practices have historically driven deforestation, which in turn exacerbates soil erosion and water scarcity.

The strong predictive power of perceived behavioral control and attitude on adoption intention suggests that policies effectively increasing these psychological drivers – through financial incentives, risk guarantees, and user‑friendly technology design – will likely accelerate the adoption of digital agricultural technologies (DAT). Higher DAT adoption rates, in turn, can reduce pressure on forest margins by enabling higher agricultural output per unit of land with fewer inputs (e.g., water, fertilizers). The significant indirect effect of perceived ease of use on intention via attitude further implies that technologies designed with minimal operational complexity will be more readily adopted, thereby maximizing the potential land‑sparing effect for forests.

Therefore, policies promoting DAT–specifically those targeting the empirical drivers identified in this study–may generate positive spillover effects for biodiversity and forest health. However, future interdisciplinary research is needed to directly quantify these cross‑sectional benefits, as our study did not measure forest outcomes. We recommend that subsequent studies link behavioral adoption models with land‑use change data to test the hypothesized conservation spillovers.

Limitations and future research suggestions

Despite the methodological rigor of the multi-stage random sampling design, several inherent limitations warrant attention. First, reliance on formal administrative divisions (province, county, district, township, and village) may overlook effective informal socio-economic layers, such as tribal networks that transcend administrative boundaries and potentially influence adoption dynamics. This may partly explain the non-significance of subjective norms. Furthermore, the assumption of intra-village homogeneity may not hold given the potential economic and agricultural diversity among farmers within a single village. The present study is cross-sectional in nature. This approach provides only an instantaneous snapshot of attitudes and tendencies at a specific point in time. Since the technology adoption process is intrinsic, dynamic, and evolutionary, and the relationships between variables may differ at various stages of awareness, experimentation, and consolidation, the findings are not able to infer definitive causality or trace the evolution of beliefs over time. Based on the findings and the limitations identified in this study, suggestions for future research are provided:

• Conducting a qualitative study with purposive sampling of farmers belonging to Tran’s boundary tribal or social networks in the region, to directly examine the influence of these informal structures on subjective norms and technology adoption.

• Conducting an intergroup study that can classify farmers from one or more villages based on socio-economic criteria (such as land size, type of farming system, access to resources) and analyzes the intergroup differences in the adoption process.

• Conducting a longitudinal study tracking the same farmers over multiple seasons to capture the dynamic evolution of attitudes and intentions as they move from awareness to trial and sustained use, revealing causal pathways over time.

• Undertaking a cross-regional comparative analysis between provinces in Iran or with other countries possessing different socio-ecological profiles. This would test the generalize ability of the current findings and help disentangle universal adoption drivers from context-specific ones.”

Conclusion

The findings of this study outline the pattern of digital agricultural technology adoption in the specific context of Western Iran and Kermanshah Province, as a key region heavily reliant on agriculture and facing unique environmental and economic challenges. The results indicate that the driving force for adoption in this region revolves primarily around rational-instrumental evaluations and feelings of individual agency, rather than social norms. This pattern is crystallized in three key findings: First, the formation of a positive attitude, which is the main driver of intention, is largely influenced by the perceived usefulness and ease of use components of TAM. This indicates that the mindset of the Kermanshah farmer, in the face of the livelihood pressures and environmental risks specific to this area, is primarily driven by the practical question, “What is the tangible and immediate benefit for me?” and “Is it possible for me to implement it with my resources and knowledge?” This finding is particularly significant given the region’s socio-institutional context. Second, the strong and significant impact of perceived behavioral control (from the TPB) on the intention to adopt emphasizes that farmers in this region make their final decision contingent on a realistic assessment of local, objective, and structural barriers, such as a lack of liquidity, limited access to technological infrastructure, or weak institutional support. Third, and in contrast, the non-significance of subjective norm is a crucial finding, deeply influenced by the region’s socio-institutional context. This result could indicate a weakening of institutional trust networks or the overriding priority of individual survival and livelihood considerations over any collective pressure or recommendation in Kermanshah province.

The integration of the TAM and TPB models for this study was not a choice, but a methodological necessity. This theoretical synthesis made it possible to simultaneously examine three layers determining adoption in this complex context: individual instrumental logic, cost-benefit and ease of use calculations (through TAM, perceived environmental constraints and barriers (through the behavioral control construct in (TPB, and the fading or absence of social pressure (through the subjective norm construct in TPB. The use of a single model, whether TAM or TPB, could not provide this comprehensive, nuanced, and contextualized picture and would fail to explain the actual reasons for farmers’ willingness or unwillingness.

Finally, while our empirical findings are specific to Kermanshah province, the integrated behavioral framework and the policy insights, particularly the limited role of subjective norms in contexts with weak institutional trust and the primacy of perceived behavioral control and attitude are likely transferable to other regions in Iran and beyond where agriculture interfaces with natural landscapes (including forests, wetlands, and rangelands), and where farmers face similar structural barriers (e.g., limited access to credit, weak extension services, and low social capital). Future cross-regional comparative studies are needed to test the generalizability of these patterns.

Author contributions

Conceptualization, A.A, A., methodology and software, M.T., Validation, A.A.A., Formal analysis, M.T., investigation and resources, A.A.A., data curation, M.T., writing original draft preparation, M.T., writing review and editing, A.A.A,. Visualization and supervision, A.A.A., All authors have read and agreed to the published version of the manuscript.

Funding

This work is based upon research funded by Iran National Science Foundation (INSF) under project No, 4024719.

Data availability

The data presented in this study are available on request from the Corresponding author.

Declarations

Competing interests

The authors declare no competing interests.