China’s remarkable economic development has undoubtedly brought about significant and tackled the challenges of sustaining and advancing a population of 1.4 billion people. However, it has also led to some unintended consequences, such as the rapid increase in production and household waste (Jiang et al. 2021; He et al. 2020). While cities have implemented more efficient waste treatment technologies, the issue of rural domestic waste management remains a pressing concern (Snellinx et al. 2021; Zhang et al. 2023). The production of domestic waste in rural areas is substantial and continuously expanding (Ma et al. 2020). According to the findings of China’s 7th census in 2021, the rural population stands at 509.79 million, with an average rural household generating 0.8 kg of domestic waste per person per day. This translates to an annual production of approximately 204 million tons of domestic waste in rural areas, growing at a rate of about 8% each year (Chen et al. 2021; Deng et al. 2022). The inadequate handling of rural domestic waste poses a severe threat to the rural ecological environment (Doyar et al. 2023). Waste disposal in rural areas primarily relies on on-site disposal, landfilling, or incineration. These methods not only hinder resourceful waste management but also contribute to local water and soil pollution, air pollution, and adversely impact the production and living environment of rural residents (Wang and You 2021). Additionally, they contribute to increased rural carbon emissions.

To address these concerns, it is crucial to focus on the separation and proper management of rural household waste. In order to enhance the management of rural household waste, China has implemented waste separation measures and introduced policies to promote waste separation in rural areas. In 2015, the Ministry of Housing and Urban-Rural Development and other relevant departments issued the “Guidance on Comprehensively Promoting Rural Waste Management” (Jiancun [2015] No. 170), which aimed to establish comprehensive waste management systems in rural areas. Subsequently, in 2021, the Ministry of Housing and Urban-Rural Development released the national standards known as the “Technical Standards for the Collection, Transportation, and Disposal of Rural Domestic Waste.” These standards provide clear guidelines and criteria for classifying and disposing of domestic waste. Furthermore, in 2022, the Ministry of Housing and Urban-Rural Development, in collaboration with six other departments, issued the Notice on Further Strengthening the Construction and Management of Rural Domestic Waste Collection and Disposal Systems (Jiancun [2022] No. 44). This notice further refined the guidelines for domestic waste disposal in rural areas by emphasizing the promotion of waste disposal systems, setting management objectives, and establishing working mechanisms.

China’s rural internet is experiencing rapid development as a result of various national policies promoting the rural revitalization strategy, including initiatives like Digital Countryside and Broadband China (Gong et al. 2021; Ma and Wang 2020). In 2021, the number of rural internet users reached 284 million, with a penetration rate of 57.6% in rural areas. This raises the question of how the widespread use of the internet in rural areas has influenced the separation and disposal of household waste. Existing research on the impact of internet use on household waste sorting has primarily focused on digital or smart technology itself, the implementation process, and the social benefits it brings (Bello et al. 2022). Some have also argued that internet data can support effective monitoring, accurate assessment, and decision-making in waste management. However, less attention has been given to how internet use affects the decision-making of individual households when it comes to choosing waste separation methods (Butler 2020).

The pervasive influence of internet technology has significantly impacted various aspects of farmers’ lives, including their environmental consciousness and decision-making processes (Huang et al. 2022). Grounded in the theory of planned behavior, the intrinsic motivation of farmers, subjective norms, and awareness of their behavior play crucial roles in shaping their involvement in environmental decision-making (Cyr et al. 2009). The internet has played a pivotal role in mitigating information disparities, granting farmers greater access to information, and reducing the cost associated with making behavioral choices. Farmers now have the means to access vital environmental knowledge through the internet, which has the clandestine effect of reshaping their values related to environmental participation. Consequently, their risk perception undergoes transformation, rendering them more economically rational in their decisions, particularly concerning waste separation (Luo et al. 2019). Consequently, the adoption of the internet may yield substantial changes in farmers’ engagement in household waste sorting practices (Kim et al. 2019).

However, there are still untapped opportunities for deeper exploration. Firstly, the majority of existing research predominantly relies on macro-level statistics or urban populations, with limited attention given to farmers and their household waste separation behaviors (Abeles et al. 2019). Secondly, prevailing studies tend to employ simple linear regression using OLS, with a scarcity of comprehensive empirical analyses that validate their mechanisms and explore their heterogeneity (Bello et al. 2022). Thirdly, while early-stage research efforts have been made (Chen et al. 2023), they primarily concentrate on arguing that environmental quality perception in the relationship between ICT use and residents’ environmental behaviors, neglecting an in-depth deconstruction of the internet’s role in farmers’ waste separation practices. Additionally, these studies adopt a broad perspective on ICT usage, overlooking the micro-level intricacies. Hence, investigating the impact of internet usage scenarios on rural residents’ domestic waste sorting becomes imperative. It serves as a vital avenue to understand how new technologies foster changes in environmental behaviors among residents, representing a valuable and meaningful research trajectory (Chen et al. 2022).

Therefore, this study investigates the influence of internet usage on rural residents’ engagement in household waste sorting within the context of emerging digital technology. Firstly, it utilizes a substantial dataset comprising 2126 households in rural China and employs the Ordered Probit Model (OPM) to empirically scrutinize the effects of internet utilization on rural residents’ involvement in waste separation. Furthermore, it discerns the nuances of these effects across different demographic groups by considering factors such as age, education, and gender, shedding light on variations in individual participation in household waste sorting due to internet use. Lastly, the study enhances the robustness of its findings by employing diverse methods, including proxy variables, alternative research techniques, and quartile analysis, in order to address potential endogeneity concerns related to the impact of internet use on rural residents’ waste separation practices. This paper makes a dual contribution:

  1. 1.

    It broadens the scope of research on farmers’ waste sorting behavior by analyzing the influence of factors such as digital technology on this behavior and precisely quantifying the marginal effects of this influence. This extension goes beyond previous studies, which primarily focused on non-new technology factors. Prior research often treated internet use as a control variable, emphasizing its impact on social and economic aspects, while giving less attention to its implications for the environment and the behavioral aspects of farmers (Reissig et al. 2022; Nogueira et al. 2022; Odhiambo 2022; Zhang et al. 2021; Yang et al. 2021). This paper delves deeper into the environmental and behavioral dimensions, thereby providing a more comprehensive and nuanced understanding of the topic.

  2. 2.

    The paper constructs a theoretical framework to elucidate the impact of digital technology, specifically internet use, on farmers’ waste sorting practices. The verified findings generally support the notion of fostering rural development, reducing carbon emissions, and advocating similar approaches in developing nations. It underscores how new technological factors have significantly influenced people’s behavior in the digital age, permeating into rural areas. Consequently, the study provides valuable insights and validation recommendations to further amplify this impact (Wu et al. 2021).

The subsequent sections are organized as follows: Section 'Literature review and research hypothesis' outlines the theoretical framework and research hypotheses regarding internet use and rural waste sorting. Section 'Model and variables' encompasses data sources, research design, methodology, and descriptive statistics. Section 'Empirical framework' presents the empirical findings and conducts analyses of heterogeneity and robustness tests concerning internet use and farmers’ domestic waste sorting. Finally, 'Robustness discussion' provides conclusions and policy recommendations.

Literature review and research hypothesis

The widespread adoption of information and communication technologies has ushered in significant changes across all facets of people’s behavior and has left an indelible mark on society at large (Zhang et al. 2020). Notably, the rapid proliferation of the internet within the digital economy has expedited this transformative process by fostering deep integration of digital technology into every sphere of people’s activities (Ma et al. 2020; Zou and Mishra 2022). An increasing number of scholars have dedicated their efforts to examining the intricate relationship between digital technologies and the tangible world (Zheng and Ma 2022; Chang et al. 2020). However, existing research has predominantly concentrated on assessing the impact of internet usage on economic growth, income augmentation, financial transactions, and overarching macroeconomic factors (Xia et al. 2022; Asongu and Le 2017). In doing so, it may have inadvertently overlooked the potential, yet equally significant, micro-level effects on individuals’ intrinsic cognitive motivations (Leng et al. 2020; Hong et al. 2017; Adeleye et al. 2021).

Economics operates on the premise of the “economic man” hypothesis, which posits that individuals make decisions aimed at maximizing personal benefits while weighing the associated costs. In the context of farmers, their behavior aligns with this hypothesis as they meticulously consider the benefits, decision costs, and transaction costs when engaging in rural waste separation (Mossberger et al. 2022). Previous research has primarily concentrated on the impact of economic factors, government subsidies, and policy advocacy on farmers’ participation in waste separation (Khan et al. 2022; Zhu et al. 2021). While these factors undeniably wield significant influence over residents’ involvement in environmental governance and waste sorting behavior, they remain within certain boundaries (Antonizzi and Smuts 2020). For instance, economic factors may only affect those who prioritize them and may not significantly sway those who are less concerned with government expenditures (Gallardo Whitacre et al. 2021). Consequently, some scholars have delved into residents’ internal perceptions as an alternative approach (Vicente-Molina et al. 2018). An increasing body of research has demonstrated that residents’ internal perceptions play a pivotal role in shaping their environmental engagement. Altering residents’ behavior necessitates the reshaping of their cognitive frameworks (Bastida et al. 2019; Chen et al. 2019). Yet, cognition is a multifaceted phenomenon, inherently subjective, and challenging to evaluate using uniform measures (Chen et al. 2016). All of these factors represent traditional influencers on rural waste sorting behavior, and the advent of new technological changes may herald a transformative shift in this paradigm.

With the rapid advancement of digital technology, internet penetration in rural areas is on the rise. Activities such as online browsing, video consumption, mobile shopping, and digital payments are steadily gaining popularity in rural China. The ubiquity of the internet in these areas has greatly enhanced people’s access to information, enabling them to access a wealth of information from around the world with just a few taps on their mobile devices (Appiah-Otoo and Song 2021). By providing accurate and comprehensive information, the internet mitigates the negative consequences of information asymmetry, thus fostering more rational decision-making. Moreover, online communication and information sharing reduce the transaction costs associated with face-to-face meetings, streamlining the decision-making process (Appiah-Otoo and Song 2021). However, it’s important to note that the internet may also expose individuals to biased beliefs and irrationality due to information pluralism (Chen et al. 2010; Chen et al. 2022).

The advantages of internet usage are manifold. Firstly, it greatly enhances residents’ access to information, allowing them to swiftly grasp the detrimental consequences of littering on the environment, their communities, and their own well-being. This knowledge leads them to prioritize longer-term choices over short-term cost savings associated with littering (Yang et al. 2021; Yang et al. 2020). Additionally, internet use facilitates knowledge acquisition and education among villagers (Zhang et al. 2023; Zhang et al. 2021). In the past, knowledge acquisition was predominantly confined to formal education settings, but the widespread availability of the internet has blurred the boundaries of information dissemination. People can now engage in online learning, particularly with advancements in digital and algorithmic technologies (Ma et al. 2020). For example, rural residents can easily access video content pertaining to local environmental protection policies, local environmental management, the adverse effects of local pollution, and more (Baker and Yang 2018). They can also access national and local information about the impact of waste classification on the environment, gradually cultivating an understanding of environmental protection and preservation (Feuerriegel et al. 2018; Doyar et al. 2023). Additionally, social networking has become increasingly prevalent among villagers and villagers, villagers and village cadres (Baker and Yang 2018; Alkhatlan et al. 2017). Platforms such as WeChat, QQ, and microblogs not only facilitate offline connections but also create online personas (Wang et al. 2015; Liu et al. 2021a; Zareie and Navimipour 2016). An online persona holds increasing influence over one’s offline reputation, with the potential to affect one’s participation in economic and social life (Hong 2017; Huang et al. 2019). Consequently, decision-makers now consider not only short-term economic factors but also the long-term implications of their online and offline presence (Zeng et al. 2019). In sum, the internet serves as a conduit for disseminating a wide range of information, including policies, regulations, and management practices to rural residents, reshaping their decision-making processes by making costs and benefits more transparent (Li et al. 2018; Li et al. 2020a).

However, internet use is not without its downsides, and research has spotlighted potential negative effects (Kallal et al. 2021). For example, psychology suggests that in the absence of objective measurement standards, the proliferation of various standards may confuse individuals when making decisions, leading them to default to their original standards or those they perceive as beneficial (Balcilar et al. 2021; Acharya et al. 2022). Some farmers in certain regions may exhibit path-dependent decision-making tendencies, making it challenging to alter their established behavioral patterns, thereby delaying rational decision-making (Alataş 2021; Lei et al. 2023). Furthermore, the internet, akin to an amplifier, has the potential to magnify both positive news and rumors. The abundance of false information and rumors in cyberspace can deter farmers from participating in waste separation and recycling (Bello et al. 2022; Salahuddin and Alam 2015). Certain individuals may be inclined to share negative or sensationalized information for attention and clicks, potentially dissuading others from participating in waste separation (Usman et al. 2021; Zhang et al. 2021).

The active promotion of new information technologies, particularly the internet, in developing countries like China, could expand the sources of information and lead to environmental spillovers that encourage greater participation in waste separation and recycling (Ma et al. 2020). However, the internet can also expose individuals to negative news and rumors, which could discourage them from participating in waste collection. The effects of internet use on waste management are multifaceted and may vary among different levels and groups of rural residents (Zheng and Walsh 2019; Hong et al. 2017). For instance, governments in these regions may pay closer attention to internet-related information and improve their management of cyberspace, potentially benefiting waste management efforts (Alkhatlan et al. 2017; Xia et al. 2022). Recent studies have delved into these intricate relationships between internet use, government policies, and waste management in developing countries (Nchofoung and Asongu 2022). Building upon this backdrop, we posit the central research hypothesis of this paper:

Research hypothesis

Internet use positively influences farmers’ participation in waste sorting behavior.

Model and variables

Model design

The Ordered Probit Model (OPM) is a suitable choice for this study as it is commonly used when assessing an individual’s choice among multiple discrete options. Since farmers in this study have four distinct options for domestic waste sorting, each with a clear ranking, the OPM is an appropriate statistical method. Following the approach of Jiang et al. (2021) and Chen et al. (2022), we employ the Ordered Probit Model to formulate the following equation:

$$y_i^ \ast = \beta _0 + \beta _1internet\,use + \mathop {\sum}\limits_{j = 2} {\beta _jcontrols_j} + \mu _i$$

In the equation, \(y_i^ \ast\) is unobservable latent variable for farmers. In this study, internetuse is the internet usage variable, the core explanatory variable. controlsi represents the control variables, encompassing the factors listed in Table 1 that may influence farmers’ household waste separation behavior. β represents the effect coefficient of each variable. Meanwhile, μi is also a noteworthy part of the above formula. It is the residual term of the model, which follows a standard positive-t distribution. Furthermore, the relationship between farmers’ domestic waste sorting behavior y and the unobservable latent variable \(y_i^ \ast\) observed through the sample is defined as follows:

$$y = \left\{ {\begin{array}{*{20}{c}} {1,} &\qquad {{{{\mathrm{y}}}} \ast \le \gamma _1} \\ {2,} & {\gamma _{{{\mathrm{1}}}} \le {{{\mathrm{y}}}} \ast \le \gamma _2} \\ {3,} & {\gamma _{{{\mathrm{2}}}} \le {{{\mathrm{y}}}} \ast \le \gamma _3} \\ {4,} &\qquad {\gamma _{{{\mathrm{3}}}} \le {{{\mathrm{y}}}} \ast } \end{array}} \right.$$

Where, γ1 < γ2 < γ3, γ1, γ2 and γ3 represent the location splitting points of farmers’ behavior towards domestic waste sorting, respectively, in order to differentiate farmers’ domestic waste sorting levels. Thus, when y* ≤ γ1, it indicates that the farmer does not participate in domestic waste separation, and y = 1. When γ1 ≤ y* ≤ γ2, it signifies that the farmer participates in sorting and selling valuable waste, assigning y = 2. When γ2 ≤ y* ≤ γ3, it means that the farmer sorts valuable waste and food waste, with y = 3. When γ3 ≤ y*, it implies that the farmer sorts valuable waste, food waste, and hazardous waste, assigning y = 4. The probabilities for the four ordered choices are expressed as:

$$\begin{array}{lll} {p(y_i^ \ast = 1\left| X \right.) = {\Phi}( {\gamma _1 - \beta _1internetuse + \mathop {\sum}\limits_{j = 4} {\beta _j} controls_j})} \\ p(y_i^ \ast = 2\left| X \right.) = {\Phi}(\gamma _2 - \beta _1internetuse + \mathop {\sum}\limits_{j = 4} {\beta _j} controls_j)\\\qquad\quad-\, {\Phi}(\gamma _1 - \beta _1internetuse + \mathop {\sum}\limits_{j = 4} {\beta _j} controls_j) \\ p(y_i^ \ast = 3\left| X \right.) = {\Phi}(\gamma _3 - \beta _1internetuse + \mathop {\sum}\limits_{j = 4} {\beta _j} controls_j)\\\qquad\quad-\, {\Phi}(\gamma _2 - \beta _1internetuse + \mathop {\sum}\limits_{j = 4} {\beta _j} controls_j)\\ {p(y_i^ \ast = 4\left| X \right.) = 1-{\Phi}(\gamma _3 - \beta _1internetuse + \mathop {\sum}\limits_{j = 4} {\beta _j} controls_j)} \end{array}$$
Table 1 Statistical frequencies of core variables.

In Eq. (3), the standard normal distribution cumulative density function is represented, and the parameters of the ordered choice model in this study will be estimated using maximum likelihood estimation.

The OPM employs the direction, significance, and magnitude of the regression coefficients to explain the dependent variable. To estimate the effects, this study uses the average marginal effect as the estimation method, relying on the direction and significance of the ordered probability regression coefficients to calculate the potential marginal effects.

Data sources

Data for this study were primarily collected through questionnaires, a valuable research tool. The questionnaire design process began in February 2020, with trial distributions conducted in selected areas during March. The data collection phase involved multiple iterations, incorporating feedback and suggestions for refinement, ultimately resulting in the finalized version of the questionnaire. Large-scale data collection occurred from April 2020 to July 2020. To ensure data reliability, we engaged a diverse group of undergraduate and some postgraduate students from Ningbo University as data collectors. Each participant received thorough training and was tasked with collecting no more than 10 questionnaires each. Our data collection efforts spanned eleven provinces across China, strategically chosen to capture variations in economic and social environments among the eastern, central, and western regions. Following a stratified sampling approach, we randomly selected several counties within each province and subsequently identified specific sample villages within each county. This meticulous process ensured a wide-ranging and diverse dataset. We received over 2200 completed questionnaires during the survey, and during the data collation phase, we identified and addressed missing values and discrepancies that did not align with actual circumstances. After these data refinement steps, we were left with a dataset comprising 2126 valid observations for analysis.

Explained variables

The primary variable of interest is farmers’ domestic waste sorting behavior (DWS). In order to measure this variable, the questionnaire included a question adapted from previous studies (Li et al. 2020): “How do you dispose of your household waste?” The response options were as follows:

  1. 1.

    “Do not sort the garbage” (DWS = 1).

  2. 2.

    “Sort and select valuable waste only” (DWS = 2).

  3. 3.

    “Sort valuable waste and food waste” (DWS = 3).

  4. 4.

    “Sort valuable waste, food waste, and hazardous waste” (DWS = 4).

Explanatory variables

In various studies, the measurement of internet use (IU) has been approached differently by scholars, leading to variations in methodologies (Liu et al. 2022). Initially, in micro-survey research, some early studies regarded internet use as a subjective perception of users and gauged it based on this criterion (Salahuddin, Alam 2015). As research has advanced, the concept of internet use has become more widely acknowledged as a variable that can be objectively measured, resulting in the development of various measurement methods (Bawden et al. 1999; Dimara and Skuras 2003). In macro-level studies, some researchers have employed indicators like the frequency of mobile phone or computer usage among a large sample of individuals (Zhang et al. 2020). However, in micro surveys, measuring internet use typically falls into two main categories: the internet adoption variable and the internet usage frequency variable.

For this study, we opted to use the internet adoption variable, which was assessed using the question: “Do you use devices such as mobile phones and computers to access information?” The choice of this question was based on our prior analysis and a review of existing research literature that employed similar phrasing, such as “Do you use mobile phones, computers, and other devices to access information?” (Zhang et al. 2017; Zhang et al. 2020). This approach is well-established in the literature and has yielded reliable results (Zhang et al. 2023; Leng et al. 2020). Respondents who answered “yes” were categorized as internet users, while those who answered “no” were classified as non-users.

To gain a more comprehensive understanding of farmers’ waste sorting behaviors and their internet usage patterns, we present a preliminary statistical table (see Table 1) with the following key observations:

  1. 1.

    Among farmers, 55.55% engage in mixed domestic waste disposal, while 44.45% opt for separate domestic waste disposal. This distribution aligns with the rural waste separation situation in China, indicating room for improvement in the implementation of rural domestic waste separation.

  2. 2.

    In the survey, 79.16% of respondents reported using the internet, whereas 20.84% stated that they do not use the internet. Notably, the proportion of internet users in our survey is slightly higher than national data on internet usage. This discrepancy may be attributed to our focus on surveying more samples in central and eastern regions, where rural internet access is more prevalent and farmer mobility remains relatively high. Our field visits in rural China affirmed the widespread use of the internet in these areas, suggesting that our survey statistics accurately reflect the prevailing conditions.

  3. 3.

    The proportion of farmers who both use the internet and engage in waste separation is notably greater than that of non-internet users who engage in waste separation. This initial observation suggests a positive correlation between internet usage and involvement in waste separation among farmers. However, it’s important to note that this conclusion will be subject to further empirical verification in subsequent analyses.

Previous research has highlighted several variables that may exert influence on farmers’ decisions to participate in waste separation, including gender, age, education, income status, identity characteristics, village characteristics, and the surrounding environment (Chen et al. 2022; Zhang et al. 2020; Liu et al. 2021b). However, while these variables are important in understanding waste separation behaviors, they are not the primary focus of this paper (Chen et al. 2021). Instead, they are included as control variables to account for the potential effects of factors other than the core explanatory variables, thereby enhancing the robustness of our findings (Chen et al. 2022; Huang et al. 2019; Liu et al. 2022).

To serve this purpose, our study collected data related to individual farmer characteristics, operational characteristics, external conditions, and geographic factors as control variables. The specific meanings and statistical values for these control variables are presented in Table 2. Notably, the statistical values for these control variables demonstrate a wide distribution of individual characteristics, aligning with the foundation of our sample analysis. These variables will be incorporated into subsequent empirical analyses to provide a comprehensive understanding of their potential impact on farmers’ waste separation behaviors.

Table 2 Decriptive statistics.

Empirical framework

Basic conclusions and marginal effect measurement

Using the level of waste sorting by farmers as the dependent variable and internet use as the core explanatory variable, the ordered probit model was applied to the survey data while controlling for other key variables. The results are presented in Table 3.

Table 3 Basic regression conclusions.

In Model (1), where only the control area is considered without the addition of control variables, the regression coefficient for internet use on farmers’ garbage sorting levels is positive and statistically significant at the 1% significance level. This finding indicates that internet use has a significant positive effect on farmers’ garbage sorting levels, providing initial evidence of the significant impact of internet use on farmers’ waste sorting behaviors. In Model (2), with the addition of control variables, the regression coefficient for internet use on farmers’ waste sorting levels decreased from 0.556 to 0.427. While the coefficient decreased, both Pseudo R-squared and LR chi2(15) values increased. This suggests that the regression findings become more robust and continue to verify the significant positive impact of internet use on farmers’ waste sorting levels.

Furthermore, an attempt was made to use OLS regression in Model (3). The results in Model (3) also demonstrate that the regression coefficients for internet use are significantly positive at the 1% significance level, reinforcing the accuracy of the conclusions drawn in Model (1). These results are generally consistent with the analysis by Zheng and Ma (2022) and align with the observations made in Table 1. They suggest that internet use indeed encourages farmers to enhance their knowledge and access to valuable information, leading them to make more environmentally favorable long-term decisions and increasing their inclination toward waste separation.

However, the ordered probit model provides limited information regarding the size of the variables’ effects. Therefore, to measure the impact of internet use on farmers’ waste sorting more accurately, this paper calculates the marginal effects of the variables, as suggested by Jiang et al. (2021).

As shown in Table 4, we measure the marginal effects of each explanatory variable under the four alternatives, with a focus on the effect of internet use on the level of waste recycling among farm households. The results indicate that compared to farmers who do not use the internet, farmers who use the internet are 14.9% less likely to not separate their household waste for disposal. Instead, they increase the percentage of waste sorted into category 2 by 3.86%, category 3 by 5.99%, and category 4 by 5.04%. This suggests that farmers who use the internet are less likely to avoid waste separation and are more inclined to choose waste separation. Notably, they have a higher likelihood of sorting waste into categories 3 and 4.

Table 4 The marginal effect of the main explanatory variable.

This data indicates that as farmers increasingly use the internet, they are more willing to engage in waste separation, thereby improving the overall level of waste separation among farmers. This change can be attributed to the increased ease of access to agricultural information through the internet, which reduces information vulnerability among farmers. It supports farmers in gaining a comprehensive understanding of the benefits of waste sorting, making them more likely to adopt this practice actively. Additionally, increased internet use may gradually influence farmers’ business philosophies through interactions with village cadres, larger village households, and relatives. Farmers become more inclined to make choices that align with their long-term interests. Moreover, as they witness the potential legal consequences of improper waste disposal through enhanced information access, farmers are more likely to make rational economic decisions, further promoting improvements in waste separation among them.

The additional control variables examined in Tables 3 and 4 reveal that indicators such as GEN, AGE, VC, NI, and EQS have insignificant effects on the extent of waste separation practiced by farm households. In contrast, DEG, PM, CFP, SC, NJS, and SEP exhibit significant impacts on the level of domestic waste separation among farm households.

Robustness discussion

To enhance the robustness of our core regression findings, we conducted various alternative analyses (Table 5). First, we explored the possibility that the choice of core explanatory variables might affect the accuracy of our regression results. To address this concern, we replaced the core explanatory variables with two alternative measures: the presence of an internet cable (“Does your home have an internet cable? Yes = 1, No = 0”) and the evaluation of home network signal quality (“Your evaluation of home network signal: Very bad = 1; Poor = 2; General = 3; Better = 4; Very good = 5”). The regression results, presented in models (1) and (2), indicate that the coefficients of the alternative explanatory variables remain significant at the 1% significance level. This suggests that our initial choice of explanatory variables does not distort the regression conclusions and reinforces the robustness of our findings.

Table 5 Robustness discussion.

Secondly, we considered the possibility that our regression method might introduce bias into the results. Therefore, we employed an alternative approach, the ordered logit model (OLM) regression. The outcomes of this analysis, presented in model (3), also show that the coefficients of the core explanatory variables remain significant, thereby confirming the robustness of our original regression findings.

Thirdly, to simplify the study further, we recoded the waste sorting variable into a binary variable: “whether the garbage is classified” (DWS1). In this binary probit model regression analysis, presented in model (4), the regression coefficients for internet use remain significantly positive at the 1% significance level. This additional analysis serves as further confirmation of the robustness of our regression conclusions.

Further discussion: heterogeneity analysis

In this section, we delve into a heterogeneity analysis to explore potential differences in the influence of internet use on participation in waste recycling behavior among various groups.

Gender differences

To investigate potential gender-based variations in the impact of internet use on household waste sorting participation, we segmented the sample by gender. The regression results, as presented in Table 6, reveal that internet use significantly increases participation in domestic waste sorting for both gender groups. Interestingly, the female group exhibits a significantly higher level of participation in domestic waste sorting after using the internet compared to the male group. This discrepancy may be attributed to the fact that, in rural China, females are typically more engaged in domestic activities, including cooking and household chores. As such, they are more likely to access information about waste separation and gain a better understanding of the environmental and ecological implications of participating in waste sorting. Consequently, they are more inclined to make rational decisions in favor of waste separation compared to their male counterparts.

Table 6 Heterogeneity analysis: Gender Differences.

Age variations

To explore potential age-related differences in participation in household waste sorting after using the internet, we categorized the sample into two age groups: those under 40 (the new generation of farmers) and those aged 40 and above (the older generation of farmers). The regression results, as shown in Table 7, indicate that a higher percentage of the older generation group participates in household waste recycling after adopting the internet compared to the new generation group. This suggests that the older generation benefits more from using the internet in terms of enriching their access to information. Older individuals may be more receptive to environmental education information conveyed through media such as short videos, which may influence their decision-making in favor of garbage recycling.

Table 7 Heterogeneity analysis: Age Differences.

Income disparities

We also examined whether different income groups in rural areas make distinct decisions regarding domestic waste sorting after adopting the internet. We divided the sample into low-income and high-income groups based on the survey’s average income level. The empirical results in Table 8 reveal that the impact coefficient of participation in household waste sorting after internet use is significantly higher in the low-income group compared to the high-income group, regardless of whether control variables were added. This suggests that the spread of the internet, particularly among rural low-income groups, can effectively bridge the information gap associated with internet use. Low-income groups have greater access to information and internet-based education, experience reduced information disparities, and encounter lower transaction costs, making it easier for them to align their decision-making with societal norms.

Table 8 Heterogeneity analysis: Net income Differences.

Educational disparities

Lastly, we explored how the educational background of farmers shapes differences in their participation in domestic waste sorting after adopting the internet. We divided the sample into two educational groups: those with less than a high school education (low education group) and those with high school education and above (high education group). The results, as presented in Table 9, consistently indicate that participation in household waste sorting after adopting the internet is significantly higher in the low-education group compared to the high education group. This disparity may arise from the fact that, for the high education group, the internet is just one of many sources of information and may not hold the same unique appeal. In contrast, for farmers in the low education group, internet use represents an incremental means of obtaining more information, affording them greater access to information rights. This, in turn, promotes more informed decision-making and facilitates their participation in household waste sorting and recycling.

Table 9 Heterogeneity analysis: Degree Differences.

These heterogeneity analyses shed light on how various demographic and socioeconomic factors intersect with the impact of internet use on farmers’ waste sorting behavior. Understanding these nuances can inform targeted strategies to encourage waste separation in rural areas.

Conclusions and discussions

This study delves into the influence of internet use on farmers’ participation in household waste sorting, providing empirical evidence to validate this impact. In the digital economy era, the internet’s widespread availability in rural areas has significantly enhanced farmers’ access to information, reduced transaction costs associated with decision-making, and lowered the learning curve for farmers (Mehmood et al. 2018; Tan and Lin 2019). Consequently, farmers are more inclined to make rational economic decisions, aligning them with policy objectives and national initiatives, particularly in the context of environmental decision-making (Zhang et al. 2017; Li et al. 2020b). This study supports the above viewpoint.

Drawing from a dataset comprising 2126 micro-surveys conducted across rural regions in eastern, central, and western China, we utilized an ordered probability model to confirm the impact of internet use on household waste sorting among farmers. The empirical analysis demonstrates that internet use substantially increases the likelihood of farmers participating in domestic waste sorting. Importantly, this finding remains robust even after subjecting it to rigorous robustness tests. Further examination through marginal analysis revealed that, compared to non-internet users, farmers who embrace the internet are 14.9% less likely to abstain from waste separation. Moreover, they exhibit a 3.86% increase in sorting waste into category 2, a 5.99% increase into category 3, and a 5.04% increase into category 4. This result is in line with the studies by Zou and Mishra (2022) and Liu et al. (2022), suggesting that new technologies, such as the internet, exert a continuous influence on the public’s information consumption and sensory experiences, which subsequently impact environmental protection and waste sorting behavior.

Furthermore, our study reveals heterogeneity in internet use among various groups regarding household waste sorting participation. We observe that internet use has a more substantial impact on rural women’s involvement in household waste sorting compared to men, echoing the findings of Leng et al. (2020) and Liu et al. (2022). Concerning age differences, internet use has a more significant effect on the participation of older generation farmers in waste segregation than their younger counterparts, in line with Liu et al. (2021b), Liu et al. (2022). Regarding income disparities, the impact of internet use on household waste sorting is more pronounced among low-income groups than high-income ones, aligning with Leng et al. (2020) and Chen et al. (2022). Finally, in terms of educational differences, internet use significantly influences participation in household waste sorting among the lower-education groups compared to the higher-education groups, echoing the findings of Liu et al. (2022) and Leng et al. (2020). This indicates that the increased adoption of the internet reduces information accessibility costs and information asymmetry risks, making it easier for marginalized groups, including women, older individuals, low-income households, and those with limited education, to benefit from the digital age’s information dividends and environmental advantages (Li and Li 2021). This finding diverges from certain existing research, notably studies by Zhang et al. (2020) and Wang et al. (2018). However, it is worth noting that there are several studies that have performed comprehensive validation, as exemplified by the works of Chen et al. (2022), Xiao et al. (2022), and Liu et al. (2022). Drawing from this finding, the literature reinforces the importance of enhancing internet utilization in the environmental sector.

In conclusion, this study not only enriches the literature on the impact of internet use on farmers’ household waste sorting but also provides a critical foundation for promoting internet use in developing countries and regions (Xue et al. 2019). In less economically developed areas, the establishment of internet infrastructure and the diffusion of digital technologies can democratize access to digital resources, enhance information value, and mitigate information asymmetry. Consequently, individuals from marginalized backgrounds, such as the poor, elderly, females, and those with limited education, can leverage the internet to access information and make decisions aligned with rational economic choices (Yin et al. 2018; Zou and Mishra 2022). This underscores the importance of the internet as a conduit for rural populations to access public goods and for governments to consider the influence of digital technology on citizens’ intrinsic perceptions and societal behavior when formulating environmental policies.

However, our study has limitations, including the use of broad survey data that may not capture fine-grained details, and the need for further investigation into the mechanisms through which internet use influences farmers’ participation in domestic waste sorting. Additionally, exploring the broader implications and value of promoting internet use in other countries warrants further research.