Application of the ‘ theory of planned behavior ’ to understand farmers ’ intentions to accept water policy options using structural equation modeling

The agricultural sector is the largest sector of water consumers, and farmers are important stakeholders involved in water conservation. This research has been conducted to determine how native farmers support different policy options to reduce agricultural water consumption. Structural equations modeling was used to construct structures derived from the ‘ theory of planned behavior ’ . For each policy option, a separate model is proposed and the modeling data supports the view that attitudes, subjective norms, and perceived behavioral control have a positive and signi ﬁ cant effect on the intention. Attitudes, and perceived behavioral control, have the strongest effect on intention. Signi ﬁ cantly, intention also have a positive impact on farmers ’ behavior. According to the results of the present research, the variance explained is over 85% for intentions and the variance explained for the farmers ’ behavior on water policy options is above 45% which is a result that indicates the high ability of the ‘ theory of planned behavior ’ in predicting policies achievement on saving agricultural water resources. It is argued that the ﬁ eld of psychology, and in particular environmental psychology, can play an important role in understanding more of the drivers to reduce agricultural water consumption and contribute to the social research program for water policy. the TPB a meta-analysis TPB, Armitage 39% in of TPB ﬁ


INTRODUCTION
One of the main reasons for food insecurity is the lack of water. Effective and integrated management of water resources has become a priority in many countries of the world. With the development of the agricultural, industrial, and service sectors, much pressure has been placed on limited water resources, while excessive use of groundwater leads to the depletion of aquifers and has significant negative consequences for natural habitats and ecosystems.
According to recent research (Dalin et al. ), Iran has ranked second in the world over the past two decades in depletion of groundwater after India. In Iran, more than 90% of water use is related to the agricultural sector and groundwater provides 62% of the water needs of irrigated lands, which have less than 30% irrigation efficiency. As a result, the agricultural sector is a vital objective for implementing water-saving policies and increasing productivity.
Developing modern irrigation techniques and technology can significantly help to control water consumption in the agricultural sector. With water-saving in the agriculture sector, water will be released for other uses. Such techniques may also increase the quality and quantity of agricultural products whilst saving water. Therefore, many national governments, by giving incentives to farmers to improve irrigation systems, are currently seeking to reduce water use in the agricultural sector, and following water-saving measures.
Over the past few decades, Iran has had many successes in dam construction and control of surface waters, and there are many investments of this sort. Environmentalists are sometimes critical that, due to exaggerated dam construction, approximately 4 billion cubic meters (4 × 10 9 m 3 ) of wetlands water rights have been allocated for other purposes. Over the past few decades, due to a self-sufficiency policy in food products, agricultural sector development policy, and increasing employment in this sector, there is a lot of pressure on water resources from the agricultural sector. According to UNSD (), using more than 40% of renewable water is considered to be severe water stress, while in Iran more than 80% of renewable water is used and so there is a 15 billion cubic meters negative balance in its plains. The total amount of renewable water in Iran has decreased from a long-term average of 110 × 10 9 m 3 to 89 × 10 9 m 3 (10-year average) (Babaee & Alijani ).
Unfortunately, the development of water infrastructure in Iran has not been balanced and not much attention was paid to increasing transmission efficiency and field efficiency along with dam constructing. Therefore, the government has been trying to increase irrigation efficiency in recent years, and for this purpose, there are many incentives for farmers to implement modern and drip irrigation. The state subsidy grants vary from 250 USD to 1,000 USD per hectare. One of the problems is the small size of agricultural land holdings due to the law of inheritance. In the event of land accumulation and implementation of drip irrigation, the government pays 85% of the cost. The government also provides educational programs for villagers to modify the pattern of cultivation and production of economical and dehydrated products and creating other jobs.
The government supports the cultivation of strategic products and guarantees their purchase. The government's environmental policies are intended to save water resources, and the present research examined the feasibility of implementing any of these policies with the 'theory of planned behavior' (TPB).
However, the question is how to increase the sensitivity of farmers to saving water and encourage them to pursue water-saving policies. Investigating factors affecting farmers' intentions and behavior can allow policymakers to target specific policies. So far, despite extensive research on this topic, empirical research on the factors affecting farmers' intentions and behaviors to adopt water policies has been very limited. Understanding farmers' incentives to adapt to water-saving policies is especially important for governments, as it increases the success of government intervention in these policies and can reduce unnecessary costs. These risks are especially important when implementing these policies in times of water scarcity. In previous research on intervention for behavior change based on De Young (), there are two approaches: 1. The 'antecedent approach': This approach follows strategies that change behavior by affecting determinants of behavior, such as commitment and information provision, and attitudinal changes. For example, (Kurz et al. ) concluded that the provision of information on water conservation has led to a 23% reduction in household water consumption.
2. The 'consequence approach': This approach follows strategies that change the behavior of the individuals by informing the positive or negative consequences of the behavior, such as giving a reward for specific behavior.
Other researchers such as Steg & Vlek (), have proposed 'information' and 'structural' approaches instead of 'antecedent' and 'consequence' approaches. The 'information' approach has the purpose of changing attitudes, beliefs, motivations, and norms. The 'structural' approach has the purpose of changing the contextual factors, such as access to products and services, regulations, or financial incentives.
In the theory of planned behavior, both approaches are considered in the components influencing the intention under the heading of attitude and behavioral control. A complete review of intervention literature goes beyond the scope of this paper. Our aim is to show how psychological research on determinants of behavior can be used to guide the development of effective interventions for implementing water policies. Our goal is to emphasize the fact that with the help of environmental psychology in the development of water policy, valuable information can be provided to inform evidence-based policy in the field of agricultural water-saving policies. Evidence-based policy (EBP) is a term often applied in multiple fields of public policy to refer to situations in which policy decisions are informed by rigorously established objective evidence. Therefore, the present study embraced the TPB (Ajzen ), which is a well know framework in psychological literature, in order to assess the determinants of farmers' behavioral intentions to implement any water resource saving policy. So, the present study could be considered as a novel approach to apply the TPB to water resources policy analysis.

STUDY AREA
Our study area is adjacent to Lake Urmia, one of the most valuable aquatic ecosystems and a UNESCO biosphere reserve ( Figure 1). Lake Urmia has been facing significant declines in water volume and area. Due to the interactions between Lake Urmia and the studied area, the importance of investigating the challenges of water-saving for sustainable development becomes more significant. Integrated assessment was made of the study area water resource by adopting the System of Environmental and Economic Accounts for Water (SEEA-WATER) and calculating over thirty water resource, economic and social indicators in 2006 and 2016, which indicated the study area is highly stressed in terms of water quantity and quality. In addition, the area suffers severely from unsustainability and dis-equilibrium between water resources and consumption (Mahdavi et al. ).

METHOD AND DATA SOURCES
Theoretical and applied foundations of the proposed model The TPB attributed to Ajzen () represents a refinement of earlier models of rational decision-making, like the found that environmental concerns were poorly significant to predict people's intent to save water. On the contrary, Chang () found that this variable has no significant effect on water-saving behaviors. Chang et al. () found that farmers' desirable attitudes toward limiting water use could predict the adoption of water-saving policies by them. Also, Tohidyan Far &

Questionnaire
The data gathering tool used was a researcher-made questionnaire designed using a literature review (Table 3). To assess the validity of the questionnaire, the content validity ratio, and content validity index were calculated using 10 expert opinions. To measure the content validity rate, experts evaluated questionnaire queries one by one on a three-point scale (1. The question is necessary; 2. The question is useful but not essential; 3. The question is not necessary) and a content validity ratio was then calculated.
To evaluate the content validity index, 10 experts were asked to specify 'relevancy', 'clarity', 'simplicity' and 'expressivity' of questionnaire queries on a 4-point ordinal scale (e.g. for 'relevancy': 1-not relevant/2-relatively relevant/3relevant/4-completely relevant) and finally, the content validity index was calculated. In this research, after collecting expert comments, all of the questions have a content validity ratio above 0.75 and a content validity index above 0.79.
Cronbach's alpha coefficient (Cr-aco) was used to examine the reliability of the set of questions related to each variable. The mean coefficient was higher than 0.7 for . Should farmers be thinking about maximizing production efficiency rather than thinking about reducing their cultivation? Q2. Do you think agricultural production more important than protecting groundwater resources? Q3. In addition to drought conditions, is it important to reduce the area under cultivation in all conditions? Q4. Do you believe that under the current conditions, the reduction of the area under cultivation for the conservation of water resources is unnecessary? Q5. Do you believe that reducing the crop area is beneficial for the conservation of water resources? Q6. Do you believe that reducing the crop area in order to protect water resources is wise? Q7. Do you believe that reducing the cultivated areas to protect water resources is your preference? Choose your answer: 1-Very little 2-Little 3-Medium 4-Much 5-Very much Social norms Q1. Most people who are important to me think that reducing the area under cultivation is considered desirable. Q2. If I reduce my cultivated area in order to save and protect the water resources, those people who are important to me will support my actions. Choose your answer: 1-Very little 2-Little 3-Medium 4-Much 5-Very much Perceived behavioral control Q1. For me, reducing the cropping areas to protect water resources is easy. Q2. If I want to, I can easily reduce the cultivation area to protect water resources. Q3. How much freedom do you have to reduce your cultivation area? Q4. Reducing the cultivated area in order to preserve water resources is an option at my disposal. Q5. How difficult is it for you to get involved in reducing the cropping areas to protect water resources? Q6. I do not have the time and skills needed to reduce the amount of cultivation to preserve water resources. Q7. For me, the application of tools to reduce the cultivation area to preserve water resources is costly. Q8. Is it possible for you to reduce the cultivation area in order to preserve water resources according to your farm conditions? Q9. It is impossible for me to carry out measures to reduce the cultivation area in order to maintain water resources. 1. Sample size is small.

Applications have a weak theoretical basis.
3. Predictive accuracy is paramount.

Correct model specification cannot be ensured.
This paper focuses on SmartPLS because it is freely available to the research community across the globe. Attitude, subjective norm, perceived behavioral control, intention, and behavior were modeled as latent reflective variables.
The crop level reduction policy conceptual model is presented in Figure 3, the crop pattern change policy conceptual model is presented in Figure 4, the policy of increasing irrigation efficiency conceptual model is presented in Figure 5, and the policy of increasing non-agricultural incomes conceptual model is presented in Figure 6.

Checking reliability and validity
To find a reliability value, the square of each of the outer loadings was verified, and all were higher than 0.70     Figure 3(b), and Behavior to BH1 in Figure 6(b), mean that the path coefficients are not significant in the above cases.

Cross-validated redundancy measures
Stone-Geisser's (Q2) values (cross-validated redundancy measures) are obtained by the blindfolding procedure in SmartPLS (Table 5). Q2 values of 0.02, 0.15, and 0.35 indicate an exogenous construct has a small, medium, and large predictive relevance for an endogenous latent variable, respectively.

Cross-validated communality measures
If cross-validated communality is positive, the measurement model has a good quality. In this research all cross-validated communality for the five variables are positive.

Determining model fit
In   if the data collected for this mode had been revised, which was accepted because the normed fit index is close to the acceptable threshold (Table 6).

Strength of the theoretical model to testing the research hypotheses
This research proposed a model that contained three independent variables and two dependent variables. The PLS algorithm was able to calculate an estimate R 2 for the dependent variable. The strength of the theoretical model was established by factor R 2 . The R 2 was calculated using the PLS algorithm with 300 iterations. Falk & Miller () suggest that the explained variance (R 2 ) should be greater than 0.1. All of the R 2 s for the four models achieved high explained variance scores. All were above the 0.  their water demand by purchasing water from the neighboring area (region 2). Therefore, region 3 has the highest water stress, and in order to adapt to this stress, farmers in region 3 intend to agree on any type of agricultural water-saving policy. They have severely reduced their land cultivation.
They also cultivated crops that are resistant to saline water and water stress (Alfalfa, Pistachio, and Elaeagnus Angustifolia). By lining of channels, they have increased the irrigation efficiency and now wish they also had non-agricultural incomes. Region '4' is mountainous and has a hard aquifer, and there are no wells in this area and the water resources are spring and qanats.
Because of the necessity for high initial capital to generate non-agricultural incomes, these incomes are limited and are usually agriculture-related such as livestock or processed crops, and a limited income from carpet weaving. Since implementation of crop level reduction policy and crop pattern change policy leads to a significant and modest reduction in agricultural incomes, it is noted that regions with very limited water resources are willing to comply with these policies. On the other hand, the implementation of increasing irrigation efficiency and increasing non-agricultural incomes policies requires high government investment.
Therefore, considering the financial facilities of the state and the intention of the farmers, the policymaker can choose the appropriate option for each region.
Regarding the high ability of intention to predict behavior, in Table 7, based on the average intentions score, prioritizations of policies to follow in the four regions and in the whole area are given.

CONCLUSIONS
Uncontrolled and inefficient use of water in the study area, especially in the agricultural sector, has increased concern and reduced water security. Therefore, the attention of researchers and policymakers to the implementation of water-saving policies has been drawn. In this study, the well-known behavioral theory of Ajzen () was used to identify the motivational factors that could encourage farmers to follow water-saving policies. Despite the widespread use of this theory in the literature on environmental behaviors, it has rarely been used to measure farmers' compliance with water-saving policies.
While past research has often focused on water conservation intentions this research moved toward measuring  It is concluded that farmers are more likely to comply with water-saving policy options when the government has provided financial incentives. Farmers will also follow policy options that will not reduce their income. In general, the economic interests of farmers are dominant in following water policy options. Many farmers think that drip irrigation reduces crop yields. They also generally have concerns about marketing low-water crops and have no experience growing such crops. Therefore, education and awareness by government agencies such as the agricultural organization can be effective. SN showed little effect on and low predictive relevance to intention in any of the four water policy option models. This shows that if farmers' attitudes to agricultural water policy are positive and that there are no barriers to doing that, they will pursue the policy without affecting social norms.
PBC had a strong influence on intentions due to the individual perception of the difficulty of implementing policy options due to financial constraints, especially in the two policy options 'Irrigation Productivity Increase Policy' and 'Non-Agricultural Income Increase Policy'. Therefore, for the implementation of these two policies, the government must provide more financial support for farmers.
ATT showed a large effect and high predictive relevance on intention in all four water policy option models. Therefore, it is recommended to design intervention programs based on promoting attitudes in order to encourage them to implement water policy.
According to the findings of this research, the explained variance over 85% for intentions and above 45% for the behavior of water policy options, suggests the high ability of the theory of planned behavior in predicting policies for saving agricultural water.

CONFLICTS OF INTEREST
The author declares no conflict of interest.