Effects of Digital Social Support on Farmers’ Behavior of Resource Utilization of Plantation Wastes

Abstract

Waste resource utilization is an important initiative in regard to promoting the development of green, low-carbon, and recycling activities in agriculture. Exploring the impact of digital social support on farmers’ waste resource utilization behavior is conducive to solving the challenges brought about by the current rapid development of the plantation industry. Based on survey data concerning 1213 farmers in Liaoning Province, this paper empirically analyzes the influence of digital social support on plantation farmers’ waste resource utilization behavior using a standard Poisson regression model and explores the roles of information comprehension and self-efficacy and the moderating role of narrow framing in the process. The results show that, firstly, digital social support had a significant positive effect on farmers’ plantation waste resource utilization behavior, and this conclusion was still valid after solving the endogenous problems. For the robustness test, every unit increase in digital social support increased the expected value of farmers’ waste resource utilization behavior by about 22.14%. Secondly, digital social support influenced farmers’ plantation waste resource utilization behavior by improving their information comprehension ability and self-efficacy. Thirdly, narrow framing will weaken the positive effect of digital social support on farmers’ plantation waste resource utilization behavior. Finally, based on the results of the heterogeneity analysis of the degree of land fragmentation and the concept of face, the influence of digital social support on the utilization of waste resources had a significant positive effect on the farmers with a low degree of land fragmentation and a strong concept of face. Therefore, the government and the agricultural sector should strengthen digital infrastructure, actively establish digital sharing platforms, and provide personalized digital social support to farmers so that they can actively engage in waste resource utilization and contribute to green and sustainable agricultural development.

1. Introduction

The resource utilization of agricultural waste is an important way to achieve the green transformation of agriculture. According to statistics, more than 3 billion pieces of pesticide packaging are discarded as waste every year in China [1], and millions of tons of agricultural film are not recycled every year [2]. The extensive use and insufficient recycling of pesticide packaging and agricultural film have led to a situation of severe environmental pollution in rural China, which hinders the development process of green, low-carbon, and recycling activities in agriculture. As waste disposers, farmers’ behavior directly affects the rural living environment. In 2024, the No. 1 Document of the Central Committee suggested “promoting the resource treatment and utilization of rural organic domestic garbage, manure, and agricultural production organic waste”. Therefore, it is of great practical significance to study farmers’ waste resource utilization behaviors and their influencing factors in order to help build a rural China with beautiful ecology and green agricultural development.

The activities of planting waste packaging treatment encompass the recycling methods of pesticide packaging and agricultural film by farmers, that is, after pesticides or agricultural film are used, farmers collect, temporarily store, and classify them, and then deliver them to recycling institutions or carry out a series of activities of reuse in specific ways. According to one study, the resource utilization of agricultural waste refers to the conversion of potentially valuable agricultural waste into renewable resource use through fertilizer, feed, energy, substrate, and industrial raw materials [3]. It has been shown that the form of agricultural waste resource utilization behavior is different in different agricultural industries, with the livestock and poultry industries mainly encompassing the resource utilization of manure [4] and the planting industry emphasizing waste packaging treatments [5] and straw resource utilization [6]. In terms of time, if the waste is properly disposed of initially, its influence will extend into the future, and this has certain intertemporal attributes.

At present, existing research on the influencing factors of farmers’ waste resource utilization behavior has mainly been carried out from three perspectives. The first of these is the individual and management characteristics of farmers. It has been found that farmers’ gender, age [7], agricultural income [8], years of farming, years of education, and planting scale [1] have an impact on farm households’ productive waste disposal methods. In terms of these factors, the improvement of farmers’ education level and agricultural income has a positive impact on farmers’ agricultural productive waste recycling behavior, while the impact of farming years on farmers’ recycling behavior is negative [9]. The second aspect is farmers’ cognitive and psychological characteristics. Studies have shown that certain factors have significant positive effects on farmers’ waste disposal behavior, such as behavioral efficacy perception [10], environmental emotion [11], environmental importance perception [12], and waste hazard perception [13]. The third aspect is external environmental factors, which can be divided in terms of formal and informal social support. The former includes factors such as policy subsidies [14], economic incentives [15], agricultural training [16], government intervention and support [17], and disciplinary supervision [18], all of which have significant positive effects on farmers’ waste resource utilization behavior. The latter includes factors such as social norms [19], social learning [20], cultural orientation [21], family support, and neighbor support [22]. All of these have significant positive effects on farmers’ pesticide packaging waste disposal behaviors.

With the advancement of the fourth technological revolution, characterized by robots and artificial intelligence, the social and economic status of human civilization is transforming into a form of digital economy [23]. In the digital age, agricultural production methods and farmers’ production behavior are inevitably affected by digital-based factors. Digital technology has been widely used in various industries and fields with strong penetration and integration, including agriculture. Digital technology plays a significant role in agriculture and can improve production efficiency, optimize resource allocation, improve the quality of agricultural products, promote sustainable development, and expand market channels, as well as using data to enhance the scientific nature of agricultural decision-making, thus promoting the development of agriculture in all aspects of modernization, intelligence, and efficiency. Farmers participate in the digital era mainly through cell phones, and the 52nd Statistical Report on the Development of the Internet in China shows that 99.8% of Chinese Internet users use cell phones to access the Internet, with the number of rural Internet users amounting to 301 million, accounting for 27.9% of Internet users as a whole. The studies that have been conducted have mainly approached the issue from the perspective of non-numerical factors, such as agricultural training [16] and government support [17]. However, little research has taken into account the changes in social support methods caused by the digital age. Therefore, this paper poses the following scientific questions: Does digital social support have an impact on farmers’ participation in agricultural resource use behavior? What are the mechanisms of influence? Combined with digital social support’s inter-temporal attributes, what role do the narrow framing characteristics of farmers play in this? These questions have not been answered.

In view of this, based on the survey data of 1213 farmers in Liaoning Province, this paper attempts to address three research objectives: Firstly, we explore the positive influence of digital social support on farmers’ waste resource utilization behavior. Secondly, we test the mediating role of information comprehension and self-efficacy. Thirdly, we test whether narrow framing plays an important moderating role in the influence of digital social support on farmers’ waste resource utilization behavior. Based on the above objectives, this paper hopes to provide effective suggestions for farmers to adopt waste resource utilization behavior.

2. Theoretical Analysis and Research Hypotheses

2.1. The Direct Impact of Digital Social Support on Farmers’ Waste Resource Utilization Behavior

Social support theory refers to the sum of behaviors of certain social networks that use certain material and spiritual means to help socially disadvantaged groups without compensation [24]. Traditional social support includes instrumental, informational, and emotional support [25]. Instrumental support is the provision of practical help and resources to meet an individual’s material needs or solve practical problems. Informational support is the provision of knowledge, advice, and information to help individuals make decisions and solve problems. Emotional support is mainly to give individuals emotional comfort, care, and encouragement to help them relieve stress and anxiety [26]. With the advent of the digital age, some or all of the traditional offline social support has migrated online. For example, farmers can obtain financial support such as loans or subsidies online through digital finance, which is one of the digital transformation methods supported by tools. Farmers can obtain agricultural related knowledge and technical information through various online methods such as SMS, WeChat, and short video platforms. According to the reality of the investigation, the digitalization degree of information support is higher than that of tools. In addition, with the popularity of smart phones, the communication mode of acquaintance society, which was mainly based on door-to-door visits, chatting, and playing cards between households, has gradually been replaced by WeChat, watching short videos, and live broadcasts, with mobile phones as the carrier. The originally offline communication has changed to online text, voice, and video, and farmers can also be encouraged or supported in online sharing and interaction, and emotional support has been partially realized online. In this paper, the above-mentioned tool support, information support, and emotional support that have been digitally transformed are collectively referred to as digital social support, which refers to an external protective measure provided by digital technology and online platforms to improve individual adaptability to the environment.

As far as waste resource utilization is concerned, farmers act as rational economic people pursuing profit maximization [27], When dealing with agricultural productive waste, they will comprehensively consider their own energy, time, factor input, etc., according to their own actual situation, and finally choose the treatment method with the lowest cost and the greatest benefit. Moreover, waste resource utilization behavior has the attribute of being a public good, the property rights of ecological environment improvement are unclear, it is impossible to charge other beneficiaries, and it is easy to have the phenomenon of a “free rider”. Farmers usually do not invest a lot of substantial costs in agricultural waste resource utilization. Digital instrumental support refers to the use of digital platforms to provide funding, subsidies, and expand sales channels for farmers, such as digital finance, e-commerce sales, etc. It can provide financial support for farmers’ agricultural waste resource utilization behaviors, thus promoting farmers to actively engage in waste resource utilization activities.

Then, digital information support can play a role by increasing the diversity of information acquisition methods and improving the scope and efficiency of information dissemination. First of all, farmers can use short video platforms, live broadcast platforms, online community communication, and other means to acquire relevant knowledge of waste resource utilization, realize the importance and necessity of waste resource utilization, and learn and master some waste resource utilization skills that can be realized without more costs. Secondly, with the help of digital devices such as mobile phones and computers, farmers can share agricultural-related videos, pictures, and texts on WeChat, Vibrato, and other software, so as to disseminate information, communicate and cooperate, share resources, and improve the efficiency of information communication.

In addition, digital emotional support mainly plays a role by improving farmers’ enthusiasm for participation and relieving stress. Farmers sharing photos or videos related to waste resource utilization may be recognized and praised by others, stimulate farmers’ sense of pride and responsibility, and improve farmers’ enthusiasm for participating in waste resource utilization. When farmers encounter setbacks or pressures in the process of waste resource utilization, they can also communicate with their families, friends, and neighbors online to receive emotional comfort and encouragement, so that they can continue to participate in waste resource utilization. Based on this, the following hypotheses are put forward.

Hypothesis 1.

Digital social support has a significant positive impact on farmers’ waste resource utilization behavior.

Hypothesis 1a.

Digital instrumental support has a significant positive impact on farmers’ waste resource utilization behavior.

Hypothesis 1b.

Digital information support has a significant positive impact on farmers’ waste resource utilization behavior.

Hypothesis 1c.

Digital emotional support has a significant positive impact on farmers’ waste resource utilization behavior.

2.2. The Mechanistic Role of Information Comprehension Ability

The influence mechanism of digital social support on farmers’ waste resource utilization behavior needs further discussion. From the aspect of digital information support, firstly, the algorithm recommendation mechanism that can be implemented on major platforms can push personalized digital information according to farmers’ individual needs, interests, and knowledge levels and greatly enhance farmers’ ability to understand and absorb information. Secondly, digital technology provides farmers with diversified information presentation methods of waste resource utilization, such as animation and video, which can visually display the processing methods, which is convenient for farmers to understand and master them. Third, through online counseling and training activities, experts and technicians are invited to provide remote guidance to farmers through video conferences and live broadcasts, and provide training videos that can be watched repeatedly, so as to strengthen farmers’ understanding of the relevant knowledge of waste resource utilization. From the aspect of digital emotional support, the digital platform provides interactive functions such as online questions and answers and forums. Farmers can ask questions about their problems on the platform, and other experienced farmers or agricultural experts can answer these questions in time. Through this interactive communication, farmers can understand information from multiple angles and solve the confusion in the process of waste resource utilization. Therefore, digital social support is helpful to improve farmers’ information comprehension ability. The research shows that digital information ability has a significant positive impact on farmers’ green production behavior, including digital information comprehension ability [28]. It can be seen that digital social support can improve farmers’ information comprehension ability and promote farmers’ participation in waste resource utilization behavior. Accordingly, the following hypothesis is proposed.

Hypothesis 2.

Digital social support promotes farmers’ information comprehension ability, and then promotes their waste resource utilization behavior.

2.3. The Mechanism Role of Self-Efficacy

Farmers’ behaviors are affected by the external environment and their own characteristics; self-efficacy refers to an individual’s perception or belief of whether he or she can adopt adaptive behaviors in the face of challenges in the environment, and studies have shown that it positively affects farmers’ green production behaviors [29]. It has also been shown that social support is effective in increasing individual self-efficacy perceptions, which in turn significantly enhances entrepreneurial behavior [30]. As far as digital social support is concerned, the information of waste resource utilization provided by digital information support can increase farmers’ relevant knowledge, skills, and experience, enhance their confidence and self-efficacy, and make farmers more inclined to undertake challenging environmental protection tasks, thus promoting farmers’ participation in waste resource utilization in the planting industry. Digital emotional support can encourage farmers, make them truly feel that their efforts and behaviors for environmental friendliness are recognized, enhance their self-efficacy, and further promote farmers’ participation in waste resource utilization. Based on this, the following hypothesis is put forward.

Hypothesis 3.

Digital society supports farmers’ self-efficacy, and then promotes their waste resource utilization behavior.

2.4. The Moderating Role of Narrow Framing

People tend to make behavioral choices based on short-term rather than long-term benefits [31]. A study found that short-sighted and far-sighted farmers have personal endowment and production decision-making differences, and far-sighted farmers often show a high degree of patience, which is more conducive to their choice of long-term investment behavior [32]. Waste resource utilization behavior is not obvious in its short-term beneficial effect, has the characteristics of benefit return lag, needs long-term persistence to produce effects on the quality of agricultural land, ecological environment, and will not bring obvious benefits to farmers in the short term, having the attribute of intertemporal benefits. Some authors believe that when people consider intertemporal problems, they often show the phenomenon of limited vision, the essence of which is the existence of people’s narrow thinking, and use narrow framing to measure the degree of people’s limited vision [33]. Narrow framing involves focusing on a specific problem or decision without considering broader contexts or related factors. In the context of digital social support for agricultural waste resource utilization, narrow framing can hinder farmers’ motivation as they may fixate on the immediate costs of digital social support rather than the long-term benefits. It can also cause farmers to selectively filter out information about digital social support that does not align with their narrow focus, preventing a comprehensive understanding. Ultimately, narrow framing may lead to sub-optimal decisions, such as rejecting digital social support due to short-term cost concerns, without fully considering its potential long-term positive impact on agricultural waste resource utilization. If farmers have a narrow frame, they will choose convenient, energy-saving, and time-saving waste disposal methods, such as throwing it directly onto the ground or burning them on the ground, instead of throwing it into trash cans or recycling it, thus weakening the positive effect of digital social support on farmers’ participation in waste resource utilization. Based on the above analysis, the following hypothesis is put forward.

Hypothesis 4.

Narrow framing weakens the positive effect of digital social support on farmers’ waste resource utilization behavior.

Based on the above analysis, this paper proposes the following research framework as shown in Figure 1.

3. Materials and Methods

3.1. Data and Sample

The data of this paper come from the grape industry research conducted by the group from July to August 2023 in Liaoning Province. The reason for taking grape growers in Liaoning Province as the research object is based on the following reasons: First, Liaoning Province has a long history of grape cultivation, and its production of table grapes ranks first in the country [34]. As a cash crop, grapes are fertilized and applied more in the production process, and in order to prevent diseases and insect pests, grapes need to be bagged, so there is more waste in the production process; Second, some small farmers do not have a strong sense of responsibility for ecological environmental protection, and their initiative and enthusiasm to participate in pollution prevention and control are not high [35]. In the process of grape production, the random discarding of pesticide packaging and plastic film waste by grape farmers often occurs, resulting in serious environmental pollution. Therefore, it is representative to select grape farmers as the research object of waste resource utilization behavior in Liaoning Province.

According to the geographical distribution of the main grape-producing areas in Liaoning Province, considering the planting scale of the region, the research group selected Jinzhou City, Yingkou City, Chaoyang City, Liaoyang City, and Shenyang City by combining stratified sampling with random sampling, and selected 18 sample towns, for each of which we randomly selected 3~4 villages, and a total of 65 sample villages. For each village, we randomly selected 10~20 grape farmers for one-on-one interviews, and obtained 1412 valid questionnaires. Combined with the content of this paper, we further eliminated the questionnaires with missing values, and finally obtained 1213 valid samples, with an effective rate of 85.91%.

The basic characteristics of grape farmers are shown in

Table 1. Basic characteristics of the respondents.

Variable	Indicators	Sample Size	Proportion/%	Variable	Indicators	Sample Size	Proportion/%
Gender	Male	793	65.38	Health status	Unhealthy	38	3.13
	Female	420	34.62		Sub-healthy	181	14.92
Age/years	≤40	132	10.88		Healthy	994	81.95
	(40, 50]	403	33.22	Number of household agricultural labor force/person	1	108	8.90
	(50, 60]	527	43.45		2	809	66.69
	>60	151	12.45		3	174	14.34
Education level	Primary and below	272	22.42		≥4	122	10.06
	Junior	819	67.52	Planting acre/hm2	≤0.33	401	33.06
	Senior	84	6.92		(0.33, 0.66]	464	38.25
	University and above	38	3.13		(0.66, 1.00]	210	17.31
Marital status	Married	1169	96.37		(1.00, 1.33]	81	6.68
	Not married	44	3.63		>1.33	57	4.70

. The sample farmers are mainly male, accounting for 65.38% of the total sample. The respondents are generally older, with more than half being over 50 years old. The education level is concentrated in junior high school and below. Healthy farmers account for 81.59%. The proportion of agricultural labor force in the families is 2, accounting for 66.69%. Most of them have small and medium-sized farms, and 88.62% of them have a planting area of 1 hectare or less. Generally speaking, the basic characteristics of the respondents are consistent with the reality of rural areas in Liaoning Province.

3.2. Model Construction

The explained variable is household waste resource utilization behavior, which is characterized by several agricultural waste resource treatment behaviors. It belongs to counting variables and can only take non-negative integers. First of all, the variance of waste resource utilization behavior is not greater than expected, and there is no excessive dispersion problem. Secondly, there are a large number of zero values in waste resource utilization behavior, which is tested by Vuong statistics. The results show that the Vuong statistics are negative, which shows that zero-expansion Poisson regression is not suitable. Therefore, this paper uses the standard Poisson regression model for parameter estimation. The model is set in the following form:

$$P(Y=y|x)={\displaystyle \frac{\exp \left[-\exp (x\beta )\right]{\left[\exp (x\beta )\right]}^y}{y!}}(y=0,1,2,3)$$

(1)

In Formula (1), $P\left(Y=y|x\right)$ denotes $\mathrm{Y}=\mathrm{y}$ as the conditional probability under the conditions of $x$, $exp\left(·\right)$ denotes the exponential function with base $e$, $\beta$ is the vector of the coefficients, $x$ is the vector of the explanatory variables, and $y^{!}$ is the factorial function.

Drawing lessons from the relevant mechanistic analysis of the research [36], we explore the transmission mechanism of digital social support on farmers’ waste resource utilization behavior from the perspectives of information comprehension and self-efficacy, as well as the moderating role of narrow framing in it, and construct the following econometric model:

$$Y_i={\gamma }_0+{\gamma }_1{X}_i+{\gamma }_2{C}_i+{\epsilon }_{1i}$$

(2)

$$Y_i={\lambda }_0+{\lambda }_1{X}_{1i}+{\lambda }_2{X}_{2i}+{\lambda }_3{X}_{3i}+{\lambda }_4{C}_i+{\epsilon }_{2i}$$

(3)

$$M_{1i}={\alpha }_0+{\alpha }_1{X}_i+{\alpha }_2{C}_i+{\epsilon }_{3i}$$

(4)

$$M_{2i}={\beta }_0+{\beta }_1{X}_i+{\beta }_2{C}_i+{\epsilon }_{4i}$$

(5)

Among them, the explained variable $Y_i$ represents $\mathrm{i}$ farmers’ waste resource utilization behavior, $X_i$ represents the core explanatory variable of farmers’ numerical social support, $X_{1i}$ represents digital instrumental support, $X_{2i}$ represents digital information support, $X_{3i}$ represents digital emotional support, $M_{1i}$ represents the information comprehension ability, $M_{2i}$ represents self-efficacy, and $C_i$ includes a series of control variables including gender, age, marital status, education level, health status, years of cultivation, whether the government has trained them or not, the number of family agricultural number of laborers, the number of plots, planting area, etc., with ${\epsilon }_1$, ${\epsilon }_2$, ${\epsilon }_3$, and ${\epsilon }_4$ as error terms. The regression coefficients of Equations (2) and (5) are represented by $\gamma$, $\lambda$, $\alpha$, and $\beta$. We focus on the magnitude and direction of the coefficients of ${\gamma }_1$, ${\lambda }_1$, ${\lambda }_2$, ${\lambda }_3$, ${\alpha }_1$, and ${\beta }_1$. Equation (2) is used to test whether there is a moderating effect of narrow framing by growers, and Equations (2) and (3) are estimated using standard Poisson regression. Equations (4) and (5) were used to test the mechanism of information comprehension and self-efficacy, estimated by least squares (OLS).

3.3. Variables

3.3.1. Explained Variable

The explanatory variable in this paper is the resource utilization behavior of farmers’ waste. Drawing on the research [37], combined with the actual situation of the grape industry, three common wastes in the grape production process, namely pesticide packaging, grape bags, and mulch, were selected for the study. The questions were, “How do you dispose of used pesticide bags/bottles?” and “How do you dispose of used mulch?” with options of 1 = throw on the ground; 2 = burn on the ground; 3 = dispose of in the garbage bin/station; 4 = recycle; 5 = other, and “What do you do with used and discarded grape bags?” with options of 1 = throw on the ground; 2 = burn on the ground; 3 = take home and burn on the fire; 4 = disposed of in the garbage bin/station; 5 = recycle; 6 = other. Since farmers do not dispose of a certain waste uniquely, drawing on the practice of existing studies [38], the largest percentage of a certain disposal behavior was used to define whether they practiced waste resource utilization or not. For example, some farmers take most of the grape bags home to burn on the fire, and a small portion is discarded in the field, so “taking home to burn on the fire” which accounts for the largest proportion is defined as the farmer’s implementation of the resource utilization of grape bag waste. Counting the disposal methods of farmers for the three types of waste, “throwing in garbage cans/stations”, “recycling”, and “taking home to burn” are defined as the farmer’s implementation of the waste resource utilization behavior, and the opposite value is assigned as 1. Finally, the explanatory variables were measured by the farmers’ participation in several waste resource utilization behaviors, taking values from 0 to 3. The statistics show that the mean value of the farmers’ waste resource utilization behaviors is 1.461, which means that on average, the farmers have less than two waste resource utilization behaviors. Specifically, 11.46% of the sample farmers did not implement any behaviors, 39.41% and 40.72% implemented one and two behaviors, respectively, and only 8.41% implemented three waste resource utilization behaviors.

3.3.2. Core Explanatory Variables

The core explanatory variable of this paper is digital social support, which is divided into digital instrumental support, digital information support and digital emotional support; combined with the aforementioned theoretical analysis and existing research [39] and data availability, we set up the question items as shown in

Table 2. Digital society support indicator system.

First-Class Index	Secondary Index	Subject Index	Mean Value	S.d	Min	Max.
Digital social support	Digital instrumental support	Do you sell agricultural products online? Yes = 1; No = 0	0.161	0.367	0.000	1.000
		Do you buy agricultural supplies online? Yes = 1; No = 0	0.371	0.483	0.000	1.000
		Do you borrow money via WeChat or Alipay? Yes = 1; No = 0	0.103	0.304	0.000	1.000
	Digital information support	How much agricultural information do you receive through your mobile phone and computer, such as technical training, market price information of agricultural products, policy information such as agricultural subsidy insurance loans, meteorological information such as rainfall, temperature, and humidity, and purchase channel information of pesticides and fertilizers or seeds?	2.481	1.586	0.000	6.000
		How many commonly used software applications like WeChat are there on mobile phones?	5.327	4.516	0.000	45.000
		How many WeChat groups are there on the mobile phone?	11.870	24.368	0.000	600.000
	Digital emotional support	Do you usually chat with people on your mobile phone or computer? Yes = 1; No = 0	0.672	0.470	0.000	1.000
		Do you share pictures, articles, or videos related to agriculture in your circle of friends? Yes = 1; No = 0	0.506	0.500	0.000	1.000
		Will your short videos be posted online? Yes = 1; No = 0	0.474	0.500	0.000	1.000

, and followed the practice of the study [40], using the entropy value method to calculate the comprehensive score of digital instrumental support, information support, emotional support, and digital social support variables, respectively.

3.3.3. Mechanistic Variables

The mechanistic variables in this paper are information comprehension ability and self-efficacy. Information comprehension ability is based on the study [41] and is measured by the question, “Can you understand the information you have found? Completely disagree = 1, Disagree = 2, General = 3, Agree = 4, Completely agree = 5”, with an average value of 4.054, indicating that farmers have a strong information comprehension ability on average. According to the previous research [42], a five-level scale was set up with questions as shown in

Table 3. Design of self-efficacy scale.

Subject Index	Mean Value	S.d	Min.	Max.
Even if some people object, you still have a way to achieve your goal.	3.748	0.952	1.000	5.000
If you try your best, you can solve most problems in your life.	4.081	0.646	1.000	5.000
You believe that you can cope with unexpected events in your life with your wisdom.	3.927	0.708	1.000	5.000
You know how to deal with unexpected things in life.	3.903	0.718	1.000	5.000
You can keep calm when encountering problems and have the ability to deal with them.	4.020	0.728	1.000	5.000
When you encounter difficulties, you can usually find many solutions.	3.990	0.695	1.000	5.000
No matter what happens to you, you can cope with it easily.	3.861	0.784	1.000	5.000
Your living environment, front and back of the house are cleaner than those of the surrounding farmers.	3.771	0.821	1.000	5.000
Compared with other farmers around you, you can bear the risk of investment failure better.	3.640	0.965	1.000	5.000
You have your own plans for your family’s annual income and how to use it.	3.918	0.911	1.000	5.000

, and finally self-efficacy was measured by the arithmetic mean of these questions, with a mean value of 3.886, indicating that on average, farmers’ self-efficacy can reach the average level.

3.3.4. Moderator Variable

The moderating variable in this paper is narrow framing, drawing on the study [33], which organizes and implements a narrow framing experiment (

Table 4. Narrow framing test.

	Decision 1	Decision 2
Question 1	A. Receive 48 game coins.	C. Lose 150 game coins.
	B. 25% chance of 200 game coins, 75% chance of 0 game coins.	D. 75% chance of losing 200 game coins, 25% chance of losing 0 game coins.
Question 2	A. Neither receive nor lose game coins.	C. Lose 100 game coins.
	B. 50% chance of losing 100 game coins, 50% chance of receiving 120 game coins.	D. 50% chance of losing 100 game coins, 50% chance of losing 0 game coins.
Question 3	A. Receive 300 game coins.	C. Lose 100 game coins.
	B. 50% chance of receiving 200 game coins, 50% chance of receiving 420 game coins.	D. 50% chance of losing 200 game coins, 50% chance of losing 0 game coins.
Question 4	A. Receive 170 game coins.	C. Lose 130 game coins.
	B. 50% chance of receiving 20 game coins, 50% chance of receiving 320 game coins.	D. 75% chance of losing 310 game coins, 50% chance of receiving 20 game coins.

) in the research to measure the presence of narrow framing among farmers.

At the beginning of the experiment, farmers were told that they had 500 game coins and needed to complete the following four questions. The goal was to have the most game coins at the end. In the experiment, farmers are faced with two decisions at the same time. In each question, if farmers can comprehensively consider the two decisions, they will calculate the number of game coins combined with AC, AD, BC, and BD, and choose the combination with the highest utility. However, if farmers have a narrow frame, they will consider the two decisions separately, that is, compare A with B, C and D, respectively, and then make a choice.

Prospect theory points out that people will avoid risks in gains and seek risks in losses. Then, taking the first topic as an example, if farmers compare A with B, they will choose A, and if they compare C with D, farmers will choose D, thus forming the AD combination. However, the income of the AD combination is a 1/4 probability of receiving 48 game coins and a 3/4 probability of losing 152 game coins, while the income of the BC combination is a 1/4 probability of receiving 50 game coins and a 3/4 probability of losing 150 game coins. Comparing the two, farmers who choose the AD combination have narrow frame characteristics, while farmers who do not have narrow frame characteristics will choose the BC combination. Therefore, the narrow frame variable of farmers who choose the AD combination is assigned as 1, and that of farmers who do not choose the AD combination is assigned as 0. The other three questions are the same.

Finally, in the four problems, if farmers show narrow frame characteristics in any one of them, the value is assigned as 1, whereas the value is assigned as 0. On the whole, 72.70% of farmers have narrow frames. In the experiment, the subjects were college students, and the results showed that 65.73% of the subjects made at least one narrow decision in the narrow framing test. This shows that the experiment in this paper is consistent with reality and the results are not biased.

3.3.5. Control Variables

Drawing on related studies [16,18,43], gender, age, health status, marital status, education level, perception of soil quality, cost of grape cultivation, years of cultivation, planting area, number of family members in agriculture, number of plots, distance from home to town (taking logarithmic values), whether they know the village rules and regulations, whether they are trained by the government, and time preference were selected as control variables. Among them, time preference draws on existing research [44], using the question “If you can receive a production subsidy from the local government, and assuming you receive CNY 1000 now, may I ask: In 1 year, do you think that this amount must be at least CNY______ in order to be the same for you as if you received CNY 1000 now?”. The discount rate is calculated according to the formula ${\mathrm{R}=(\mathrm{F}/\mathrm{P})}^{(1/\mathrm{t})}-1$, where $P$ represents the value now, $F$ represents the value of the extension, $t$ represents the extended time (12 months), and $R$ represents the discount rate, which ultimately represents the degree of time preference of the farmer, where the smaller the discount rate, the greater the degree of time preference.

The specific descriptive statistics of the above variables are shown in

Table 5. Descriptive statistics of variables.

Variable Name	Explanation and Assignment	Mean Value	Sd.	Min.	Max.
Waste resource utilization behavior	How many kinds of waste resource utilization behaviors do farmers actually participate in? Do not participate = 0; Participate in 1 kind = 1; Participate in 2 kinds = 2; Participate in 3 kinds = 3	1.461	0.804	0.000	3.000
Digital social support	Entropy synthesis	0.343	0.123	0.000	1.038
Digital instrumental support	Entropy synthesis	0.065	0.076	0.000	0.305
Digital information support	Entropy synthesis	0.017	0.008	0.000	0.086
Digital emotional support	Entropy synthesis	0.091	0.058	0.000	0.166
Information comprehension ability	Can you understand the information you have found? Completely disagree = 1; Disagree = 2; General = 3; Agree = 4; Completely agree = 5	4.054	0.793	1.000	5.000
Sense of self-efficacy	Arithmetic mean	3.886	0.489	2.000	5.000
Narrow frame view	Yes = 1; No = 0	0.727	0.446	0.000	1.000
Gender	Male = 1; Female = 0	0.654	0.476	0.000	1.000
Age	Age of interviewee/years	51.169	8.050	22.000	65.000
Health status	Unhealthy = 1; Sub-healthy = 2; Healthy = 3	2.788	0.479	1.000	3.000
Marital status	Married = 1; Not married = 0	0.964	0.187	0.000	1.000
Education level	Primary school and below = 1; Secondary school = 2; High school = 3; University or above = 4	1.908	0.641	1.000	4.000
Cognition of soil quality	Score the soil fertility or grape suitability of the largest plot in your home, 1–10 points. How many points do you rate?	7.627	1.633	1.000	10.000
Grape planting cost	Grape planting cost/CNY 10,000	3.539	12.009	0.050	400.000
Planting years	Grape planting time/year	21.108	9.560	1.000	50.000
Planting area	Grape planting acre/hm2	0.669	1.296	0.033	33.333
Family agricultural population	Number of agricultural laborers in the family/person	2.256	0.755	1.000	4.000
Number of plots	Number of grape planting plots/plots	2.545	1.357	1.000	5.000
Distance from home to town	Actual distance from home to town/km (take logarithm)	1.431	0.537	0.000	3.045
Do you know the village rules and regulations	Yes = 1; No = 0	0.528	0.499	0.000	1.000
Does the government train	Yes = 1; No = 0	0.629	0.483	0.000	1.000
Time preference	Discount rate	0.211	0.302	0.001	0.750
Investigation area (Jinzhou City as control group)	Yingkou City = 1; Others = 0	0.338	0.473	0.000	1.000
	Chaoyang City = 1; Others = 0	0.064	0.245	0.000	1.000
	Shenyang = 1; Others = 0	0.071	0.257	0.000	1.000
	Liaoyang City = 1; Others = 0	0.058	0.233	0.000	1.000

.

4. Results

4.1. Model Estimation

Table 6. Regression results.

	Model (1)	Model (2)	Model (3)	Model (4)
Digital social support	0.614 *** (4.078)	0.351 ** (2.205)
Digital instrumental support			0.443 ** (2.069)	0.177 (0.795)
Digital information support			1.419 (0.778)	0.1678 (0.090)
Digital emotional support			1.222 *** (2.720)	0.910 ** (2.036)
Gender		0.0627 * (1.914)		0.063 * (1.920)
Age		−0.002(−0.744)		−0.001(−0.275)
Health status		0.069 ** (1.978)		0.077 ** (2.160)
Marital status		0.027(0.358)		0.041 (0.561)
Education level		0.122 * (1.825)		0.121 * (1.812)
Cognition of soil quality		0.015 (1.521)		0.016 (1.544)
Grape input cost		0.003 ** (2.344)		0.003 ** (2.379)
Number of family agricultural labor force		0.014 (0.661)		0.017 (0.765)
Planting years		0.003 (1.277)		0.001 (0.638)
Planting area		−0.001 (−1.478)		−0.001 (−1.545)
Number of plots		0.024 * (1.774)		0.026 * (1.938)
Distance from home to town		−0.014 (−0.472)		−0.016 (−0.524)
Do you know the village rules and regulations		0.037 (1.130)		0.037 (1.125)
Does the government train		0.039 (1.143)		0.039 (1.125)
Time preference		0.022 (0.498)		0.028 (0.545)
Investigation area		Controlled		Controlled
Constant term	0.210 *** (4.661)	−0.126 (−0.642)	0.244 *** (6.500)	−0.149 (−0.856)
Wald Chi2	16.630	128.080	22.820	112.860
Prob > (Chi2)	0.000	0.000	0.000	0.000
Pseudo R2	0.003	0.014	0.003	0.013
N	1213

Note: (1) *, **, and *** indicate significance at the 10%, 5%, and 1% levels of significance, respectively. (2) Figures in brackets are standard errors.

reports the regression results based on the Poisson regression model. Among them, model (1) and model (3) did not include control variables, model (2) and model (4) included control variables, and the Wald chi-square values all passed the significance test at the 1% statistical level. The coefficient of digital social support in model (1) is significantly positive at the 1% level, indicating that digital social support has a significant positive impact on farmers’ waste resource utilization behavior. However, failure to include control variables leads to endogenous bias, which biases the model estimates to be inconsistent, interferes with causality judgments, and reduces the explanatory power of the model. Therefore, we added control variables to model (2). The coefficient of digital social support is also significantly positive, indicating that digital social support has a significant positive impact on farmers’ waste resource utilization behavior, and hypothesis H1 is verified. By finding the marginal effect, it can be seen that the coefficient of digital social support is 0.200, which indicates that for every unit increase in numerical social support, the expected value of farmers’ waste resource utilization behavior increases by about 22.14%.

Models (3) and (4) reported the effects of three sub-dimension variables of digital social support on farmers’ waste resource utilization behavior. The results show that hypothesis H1a is not verified. Because of the results of model (4) with the inclusion of control variables, the coefficient of digital instrumental support is not significant. This is probably because it is costly for farmers to use digital tools to support them, that is, to use digital platforms to obtain funds or other support. Secondly, digital emotional support has a significant positive effect on farmers’ waste disposal behavior, indicating that farmers communicate with other people through digital platforms, and the attention, encouragement, and support of others promote farmers to regulate their own behaviors and promote them to adopt waste resource utilization behaviors, and hypothesis H1c is verified. Thirdly, the coefficient of digital information support is not significant and hypothesis H1b is not tested. There are two possible reasons, one is that the education level and digital literacy of farmers are uneven, and some of them are weak in accepting and applying digital technology; the other is that traditional waste disposal habits formed by farmers over a long period of time, such as casual disposal and burning, are difficult to change fundamentally in the short term as a result of the influence of digital-based information.

4.2. Robustness Test

Combined with the data characteristics and attributes of the explained variables, the robustness test is carried out in the following ways. Firstly, the least squares (OLS) model is applied for estimation, and the regression results are shown in model (1) and model (2) in

Table 7. Estimation results of robustness test.

	OLS (Transformation Model)		Oprobit (Transformation Model)		Poisson (Excluding Samples over 60 Years Old)
	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)
Digital social support	0.509 **		0.719 **		0.418 **
	(2.216)		(2.230)		(2.359)
Digital instrumental support		0.275		0.392		0.177
		(0.798)		(0.813)		(0.771)
Digital information support		0.198		0.312		0.729
		(0.073)		(0.082)		(0.354)
Digital emotional support		1.343 **		1.881 **		1.028 **
		(2.045)		(2.050)		(2.104)
Control variables	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Constant term	0.769 ***	0.701 **			−0.127	−0.157
	(2.772)	(2.501)			(−0.594)	(−0.724)
N	1213	1213	1213	1213	1062	1062
R2	0.069	0.079	0.038	0.034	0.016	0.014

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

; the coefficient of digital social support is significantly positive, which indicates that digital social support has a significant positive impact on farmers’ waste resource utilization behavior, and the coefficient of digital emotional support is also significantly positive, which indicates that digital emotional support positively affects farmers’ waste resource utilization behavior, consistent with the results of the benchmark regression. Secondly, the ordered multiple regression model (Oprobit) was applied for estimation, and the results are shown in model (3) and model (4) in Table 7, which also show that the coefficients of digital social support and digital emotional support are significantly positive, consistent with the results of the benchmark regression. Finally, since older farmers are less familiar with the use of digital devices such as cell phones, which may bias the results, the samples aged 60 years or older are excluded, and the samples aged less than or equal to 60 years are retained, and the results, as shown in model (5) and model (6) in Table 7, also demonstrate that the coefficients of digital social support and digital emotional support are significant and positive at the 5% significance level, which indicates that the coefficients of digital social support and numerical emotional support coefficients have a significant positive effect on farmers’ waste resource utilization behavior. The above tests show that the results of this paper are robust.

4.3. Endogenous Discussion

Omitted variables, measurement errors, reverse causation, and sample selection problems may all lead to endogeneity problems, making the model estimates inaccurate. The inconsistency of network conditions between rural villages may result in sample self-selection for access to digital social support by farm households, causing endogeneity problems. In addition, variables that are unobservable and can affect the behavior of waste resource utilization, such as the moral level and personality of farmers, are not directly accessible, and there are omitted variables, causing endogeneity problems. In this paper, the instrumental variable method is applied to test endogeneity.

The instrumental variable of digital social support is selected as the “availability of digital intelligent machinery”. Instrumental variables are required to have the characteristics of being related to the independent variable and not directly related to the dependent variable. If farmers have digital intelligent mechanical equipment in their homes, such as water and fertilizer integrated machines with digital devices, rolling machines, intelligent temperature control, fan, automatic light replenishment machines, drones, etc., it indicates that their digital social support is high and meets the relevance of instrumental variables, and whether or not they have intelligent mechanical equipment in their homes is not directly related to their participation in the behavior of waste resource utilization. Waste resource utilization behavior has no direct relationship, satisfying the exogeneity of instrumental variables.

The estimation was carried out using the 2SRI (Two-Stage Residual Inclusion) method, which is a method used to solve the endogeneity problem of nonlinear models, and the regression results are shown in

Table 8. Endogenous issues test.

	OLS	Poisson
	Digital Social Support	Waste Resource Utilization Behavior
Digital social support		4.722 ***
		(2.602)
Digital intelligent mechanical equipment or not	0.019 ***
	(2.768)
Control variable	Controlled	Controlled
Constant term	0.275 ***	−1.449 **
	(7.238)	(−2.491)
N	1213	1213
R2	0.192	0.015
First-stage residual		0.324 **
		(2.019)
F-value in the first stage	15.650
Wald chi2	139.730

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

. The results show that the coefficient of digital social support is positive at the 1% significance level, and the first-stage residuals are positively significant, while the first-stage F-value is greater than the empirical reference value of 10, indicating that there is no problem of weak instrumental variables, indicating that the digital social support has a significant and positive promotion effect on the behavior of resource utilization of farmers’ waste, which is in line with the results of the baseline regression, and verifies the validity of the results.

4.4. Mechanistic Analysis

Further testing of the mechanism of the influence of digital social support on the behavior of farmers’ waste resource utilization is verified using the method of the study [36], and the regression results are shown in

Table 9. Mechanism analysis.

	OLS		Poisson Waste Resource Utilization Behavior
	Information Comprehension Ability	Sense of Self-Efficacy	Narrow Frame	No Narrow Frame View
	Model (1)	Model (2)	Model (3)	Model (4)
Digital social support	1.265 ***	0.695 ***	0.224	0.569 **
	(5.508)	(5.219)	(1.191)	(1.966)
Control variable	Controlled	Controlled	Controlled	Controlled
Constant term	2.825 ***	2.533 ***	−0.066	−0.109
	(9.388)	(13.867)	(−0.301)	(−0.250)
N	1213	1213	882	331
R2	0.075	0.088	0.012	0.032

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

. Model (1) in Table 9 shows that the coefficient of the impact of digital social support on the information comprehension ability of farmers is 1.265, and it is significantly positive at the 1% significance level, which indicates that digital social support can significantly enhance the information comprehension ability of farmers, and since the information ability can promote the adoption of waste disposal behavior by farmers [45], it can be assumed that digital social support, by enhancing the information comprehension ability of farmers, thus promotes their participation in waste resource utilization behavior. Hypothesis H2 is verified.

Similarly, the coefficient of digital social support in model (2) in Table 9 is 0.695, which is significantly positive at the 1% level, indicating that digital social support significantly improves farmers’ self-efficacy, and it has been shown that self-efficacy promotes farmers’ behavior of waste resource utilization [46], so digital social support promotes farmers’ self-efficacy through enhancing their waste resource utilization behavior. Hypothesis H3 was verified.

In addition, to test the moderating role of narrow framing, the whole sample was grouped by the presence of narrow framing in farmers as a grouping criterion and estimated separately, and the results are shown in model (3) and model (4) in Table 9. The results of model (4) show that the coefficient of digital social support is significantly positive at the 5% level among farmers without the presence of narrow framing, indicating that digital social support significantly and positively affects the behavior of farmers’ waste resource utilization. While in the farmers with narrow framing, as shown in model (3), the positive facilitating effect of digital social support on farmers’ waste resource utilization behavior disappeared, indicating that narrow framing weakened the positive effect of digital social support on farmers’ waste resource utilization behavior, and hypothesis H4 was verified. It can be seen that even if farmers with narrow framing receive numerical social support, the probability of adopting waste resource utilization behavior is lower because they do not value long-term benefits.

4.5. Heterogeneity Analysis

With different degrees of land fragmentation, farmers have different operation modes and costs of waste treatment. Land fragmentation makes it difficult to carry out waste resource utilization on a large scale, it is more difficult to operate, and carrying out waste resource utilization requires more time and labor costs, which farmers may not consider cost-effective. There are cluster differences in the effect of digital social support on the waste resource utilization behavior among groups with different degrees of land fragmentation. Therefore, this paper, based on existing studies [47] and combined with the actual situation of the research, measured the degree of land fine-fragmentation by the average land plot area (cultivated area/number of plots), divided into two groups of farmers with the mean value as the division, and estimated them separately; the results are as shown in

Table 10. Heterogeneity analysis of land fragmentation degree and face concept.

	The Degree of Land Fragmentation Is High	The Degree of Land Fragmentation Is Low	Strong Sense of Face	Weak Concept of Face
	Model (1)	Model (2)	Model (3)	Model (4)
Digital social support	0.154	0.542 *	0.516 **	0.128
	(0.812)	(1.895)	(2.355)	(0.573)
Control variable	Controlled	Controlled	Controlled	Controlled
Constant term	−0.472 *	0.611 *	(−1.906)	−0.236
	(−1.914)	(1.960)	−0.076	(−0.865)
N	850	363	591	622
R2	0.017	0.023	0.018	0.018

Note: * and ** indicate significance at the 10% and 5% levels of significance, respectively.

for model (1) and model (2). According to the results, it can be seen that the effect of digital social support on waste resource utilization behavior is not significant in farmers with a high degree of land fragmentation, while in farmers with a low degree of land fragmentation, digital social support has a significant positive effect on waste resource utilization behavior.

Farmers’ face concept may lead to differences in their waste disposal behavior, therefore, using the criterion of “whether farmers care about other people’s views and evaluations” [48], farmers were divided into two groups with strong and weak face concepts, and the results were estimated separately. The results are shown in model (3) and model (4) in Table 10. Digital social support has a significant positive effect on farmers’ waste resource utilization behavior in the sample with a strong face concept, while it is not significant in the farmers with a weak face concept. The possible reason is that for farmers with a strong face concept, they pay more attention to their reputation and image in digital social support, do not litter planting waste, and keep their vineyards clean and tidy, which is conducive to obtaining more appreciation from others when sharing pictures and videos of their own planting process, and their self-behavior is encouraged, and their waste resource utilization behavior is strengthened.

5. Discussion

5.1. Research Conclusions

Based on the survey data of 1213 grape farmers in Liaoning Province in 2023, this paper applies the Poisson regression model to explore the influence of digital social support on farmers’ behavior of plantation waste resource utilization and its mechanism. The following conclusions are obtained: Firstly, digital social support significantly promotes farmers’ adoption of waste resource utilization behaviors, in which digital emotional support has a significant positive effect, while digital informational support is not significant, and the conclusions are still valid after solving the endogenous problems by using the 2RIS model, as well as replacing the estimation method and reducing the regression samples for the robustness test. Secondly, the mechanistic test illustrates that digital social support promotes farmers’ waste resource utilization behavior by improving their information comprehension and self-efficacy. Thirdly, the presence or absence of narrow framing in farmers plays a moderating role in the effect of digital social support on their waste resource utilization behavior, i.e., narrow framing weakens the positive effect of digital social support on farmers’ waste resource utilization behavior. Fourthly, the heterogeneity analysis shows that the influence of digital social support on farmers’ waste resource utilization behavior is significant in the group with low land fragmentation and a strong face concept.

The results of this study have far-reaching implications in many aspects. At the social level, digital social support, especially digital emotional support, can improve farmers’ information understanding and self-efficacy, encourage farmers to actively participate in the utilization of waste resources, and help to change the concept of rural society and promote the formation of green production and lifestyles. On the environmental level, it significantly promotes the behavior of waste resource utilization, helps to reduce agricultural waste pollution, increases the utilization rate of resource recycling, and improves the rural ecological environment. At the economic level, it can help farmers to reduce costs, improve production efficiency, and promote sustainable agricultural development, and its impact is more significant among specific groups, providing a reference for the precise formulation of agricultural support policies and contributing to the revitalization of the countryside and the long-term development of the economy.

5.2. Marginal Contribution

Compared with previous studies, the marginal contribution of this research lies in the following three points. Firstly, current research suggests that digital technologies offer effective solutions for reducing resource waste [49]. Therefore, we construct a set of digital context-based metrics of digital social support, focusing on how such support affects farmers’ agricultural waste resource utilization behavior. Secondly, previous studies have shown that agricultural training [16] and government intervention and support [17] promote the adoption of waste resource utilization behaviors by farmers, and our study is consistent with the results of previous studies. However, based on existing studies, we broadened our research perspective by considering the digital era and including digital-based emotional support in our analysis, which was found to have a significant facilitating effect. In addition, some studies have found that social learning affects farmers’ behaviors [20]. However, in this paper, the role of digital-type information support is not significant, which is different from previous findings. Thirdly, we explored the moderating role of narrow framing in this relationship. We aimed to uncover the conditions under which digital social support is most effective in promoting sustainable agricultural practices. This study contributes to the understanding of how digital technologies can be leveraged to enhance resource utilization behaviors in rural contexts, while also highlighting the importance of psychological and cognitive factors such as narrow framing in shaping farmers’ decision-making processes.

5.3. Policy Recommendations

Based on the above conclusions, the policy implications of this paper are as follows: Firstly, agricultural departments and institutions should partner with private entities and technology developers to create digital platforms offering social support. This will boost farmers’ environmental awareness and involvement in waste resource utilization behavior. Additionally, the government must improve digital infrastructure in collaboration with private firms and non-governmental organizations (NGOs) to deliver digital social support. This support will enhance farmers’ information processing skills and self-efficacy, further driving their participation in waste resource utilization behavior. Secondly, it is necessary to increase the radius of digital social support, so as to encourage farmers to recognize the importance and necessity of waste resource utilization by external outreach, training, or funding, promote their pursuit of long-term benefits, and mitigate the inhibiting effect of narrow framing in the process. Thirdly, personalized digital social support should be provided to farmers. Digital platforms and software will be used to hold events for sharing and exchanging information on agricultural production processes, provide information on methods and benefits of large-scale waste resource utilization to farmers with a low degree of land fragmentation, and provide opportunities for face-saving farmers to demonstrate and promote demonstration effects, and take into full consideration the roles of local community leaders or cooperatives in terms of emotional support and trust in this regard, so as to promote the active participation of farmers in waste resource utilization and contribute to the construction of green and beautiful villages.

5.4. Limitations and Future Research

This study takes the social support theory and the theory of farmers’ behavior as the basic framework structure to investigate the influence of digital social support on farmers’ waste resource utilization behavior. This paper has the following limitations: Firstly, the data in this study only come from a study of grape growers in the main production area of Liaoning Province, and in the future, the study area can be expanded to the northeastern region or even the whole country to improve the representativeness of the data. Secondly, this study only used cross-sectional data, and panel data can be used in the future in order to observe the dynamic impact of digital social support on farmers’ waste resource utilization behavior. Finally, the research perspective can be further broadened, and in the future, we can compare the impacts of digital and non-digital social support on farmers’ waste resource utilization behaviors, and answer the question of which one has a greater impact, so as to put forward policy suggestions for the green and sustainable development of agriculture.