Influential Factors Affecting Recycling Behavior toward Cardboard Boxes in the Logistics Sector: An Empirical Analysis from China

Abstract

With the escalating issues of resource waste and environmental pollution, the effective recycling of cardboard boxes within the logistics sector has emerged as a crucial factor in advancing sustainable development. This study employs the extended theory of planned behavior (ETPB) to devise a questionnaire and gather data from 700 respondents in China, aiming to analyze the influential factors that impact consumers’ engagement in recycling mechanisms provided by express delivery companies. Utilizing a principal component analysis, five co-factors that influence consumers’ willingness to recycle are identified. The findings of a multinomial logistic regression reveal a positive correlation between these five co-factors and recycling behavior, with attitude exhibiting the greatest significant influence (5.076 times in model 1 and 2.375 times in model 2) on recycling behavior. These results will serve as a scientific foundation for express delivery companies and governmental entities to adapt and optimize existing environmental protection policies, thus fostering sustainable development.

1. Introduction

The advancement of economic development and the transformation of consumer consumption patterns have caused e-commerce to emerge as the preferred shopping method [1], resulting in an increased demand for the transportation of goods. In 2022, the express delivery industry witnessed staggering growth, with parcel volume reaching 110.58 billion pieces and generating USD 14.8 billion in revenue [2]. According to China’s 14th Five-Year Plan, it is projected that the annual revenue of the express delivery sector will exceed CNY 1.8 trillion, and the parcel volume will surpass 150 billion pieces by 2025 [3]. However, this significant surge in express delivery volume will inevitably lead to a substantial increase in packaging waste [4].

Within the logistics sector, packaging not only serves as a means of convenient transportation but also plays a crucial role in protecting goods during transit [5]. Cardboard boxes, commonly used as packaging materials, are derived from fiber-based materials, such as wood [6], which are natural resources. However, the consumption of natural resources has witnessed a significant, more than threefold increase in the past 50 years [7]. This has led to a dilemma in which natural resources are both insufficient and being depleted, as noted by Hussain et al. (2020) [8]. Enhancing the efficiency of natural resource utilization represents the most effective approach to mitigating the adverse environmental impacts resulting from overexploitation [9].

Cardboard boxes possess properties such as availability, recyclability, and biodegradability [10,11,12,13], making them ideal candidates for resource utilization via recycling at the end of their life cycle [14]. Unfortunately, a substantial number of recyclable materials, including cardboard boxes, find their way into municipal solid waste without undergoing appropriate treatment [15]. Consequently, many communities and organizations are taking up the mantle of social responsibility by establishing recycling mechanisms to address this issue and improve the current situation.

In 2019, various express delivery companies in China, including STO Express, Yunda Express, YTO Express, and others, made a collective announcement to deploy 50,000 green recycling boxes nationwide. These recycling boxes enable consumers to conveniently deposit their packaging wastes, which primarily comprises cardboard boxes. Despite this initiative, consumer participation in recycling activities remains low. Thus, from the demand perspective, this research aims to examine the influential factors affecting consumers’ recycling behavior with respect to recycling cardboard boxes under the mechanism established by express delivery companies. The findings of this study hold practical significance for enhancing the efficiency of recycling resources within the logistics sector.

Fan et al. (2017) [16] demonstrated that the generation of packaging waste by the express delivery industry creates a significant environmental burden in China. Consequently, the management of packaging waste has become a prominent area of research interest. From an industrial perspective, Kremser et al. (2022) [17] proposed an innovative recycling procedure to maximize material recovery. Guo et al. (2022) [18] developed an evaluation model based on a life cycle assessment (LCA) to minimize the environmental impact during the design stage of recyclable express boxes. Shi et al. (2023) [19] established a circular symbiosis network to recycle express packaging waste and put forth practical recommendations for express companies. While studies in this field have explored the management of express packaging waste, including cardboard boxes, which possess significant recycling potential, there is a paucity of detailed research focusing on specific recycling modes.

The study of consumer recycling behavior holds significant practical value as it can contribute to the promotion of resource utilization efficiency. To explore this area, researchers have investigated various aspects relating to consumers’ recycling behaviors. For instance, Chen et al. (2018) [20] examined the influence of ecological perceptions on consumers’ willingness to recycle beverage packaging. Dong et al. (2018) [21] analyzed the factors affecting consumers’ willingness to recycle express packaging waste using a neural network approach. In a different context, Yang et al. (2021) [22] developed a tripartite game model involving governments, consumers, and take-out platforms to analyze preferences for participating in the recycling of take-out packaging waste. These studies collectively contribute to the understanding of consumer recycling behavior and offer valuable insights for practical applications.

This study aims to address two research questions related to the existing low recovery efficiency in the recycling mechanisms established by express delivery companies. The first research question investigates the factors that influence consumer participation in these recycling mechanisms, while the second question explores the policies implemented by governments and companies to promote maximum consumer engagement in recycling activities. To tackle these research questions, we designed questionnaires based on the extended theory of planned behavior (ETPB) and collected data from 700 respondents in China. A principal component analysis was employed to identify five common factors that affect consumer behavior. Through a multinomial logistic regression analysis, we found that in terms of consumers’ willingness and reluctance to participate in recycling activities, the factors ranked in order of importance are attitude, subjective norms, benefits, and intention (model 1). Additionally, when considering the choice of behavior between being willing to participate and indifferent to participating in recycling activities, attitude, subjective norms, benefits, and perceived behavioral control sequentially influence consumer recycling behavior (model 2).

In summary, our research offers several key contributions. Firstly, we introduce the extended theory of planned behavior (ETPB), which builds upon the traditional theory of planned behavior (TPB) and is particularly relevant to the study of recycling behavior. This expansion enhances the theoretical framework for behavior research in this context. Secondly, the findings of this study can serve as valuable guidance for relevant departments, enabling them to implement effective and targeted measures aimed at enhancing the efficiency of recycling initiatives.

The structure of this paper is organized as follows: Section 2 presents the theoretical model and provides an overview of the current recycling system established by enterprises. Section 3 briefly introduces the research design, data source, and multinomial logistic regression model used in the study. In Section 4, we explain the results of the analysis and provide a detailed discussion. Lastly, Section 5 comprises relevant practical recommendations, the limitations of this study, and directions for future research.

2. Theoretical and Practical Background

2.1. Theoretical Analysis

The theory of planned behavior (TPB), originally proposed by Ajzen, is a well-established social psychological theory that examines individual behavior [23]. The TPB posits that behavior is influenced not only by individuals’ intentions but also their abilities and the external factors that enable them to engage in specific behaviors [24]. The theory comprises three key elements: the attitude toward the behavior, subjective norms, and perceived behavioral control. The TPB suggests that individuals’ behaviors are directly influenced by their intentions, which, in turn, are shaped by the combined effects of these three variables. Scholars have employed the TPB as a theoretical foundation when assessing consumers’ intentions toward recycling behavior. For instance, Ding et al. (2023) [25] explored the determinants of contractors’ recycling intentions toward construction and demolition waste using the TPB. Wan et al. (2021) [26] investigated residents’ recycling intentions by incorporating place attachment into an extended theory of planned behavior. Lou et al. (2022) [27] augmented the TPB with recycling convenience to examine the intention of electric bike users to recycle waste lead–acid batteries. In this research, we introduce recycling benefits into the TPB to examine consumers’ preferences for recycling cardboard boxes within the recycling mechanism provided by delivery companies.

2.1.1. Attitude toward Recycling Behavior

In the theory of planned behavior (TPB), attitude refers to an individual’s inclination or willingness to engage in a specific behavior, and it is considered a crucial factor influencing the likelihood of behavior enactment. Numerous studies have examined individuals’ attitudes toward their behaviors across various domains. For instance, Escario et al. (2020) [28] employed a logistic regression analysis to explore the relationship between environmental attitudes and waste-related behaviors, utilizing a sample of Spanish residents. Their findings demonstrated a positive association between environmental attitudes and waste-related behaviors. Similarly, Cai et al. (2021) [29] employed the conditional value method (CVM) to assess differences in attitudes and willingness to pay among individuals, specifically regarding express packaging. Their study revealed that people’s knowledge of express packaging is linked to their behavioral choices. Based on these studies, we propose our first hypothesis:

Hypothesis 1 (H1).

The greater the strength of their attitude toward recycling cardboard boxes within the recycling mechanism provided by express delivery companies, the more likely individuals are to engage in recycling behavior.

2.1.2. Subjective Norms

Subjective norms refer to individuals’ perceptions of the support or opposition they receive from significant others, such as parents, friends, and the general public, regarding a specific behavior. The influence of subjective norms on individual behavior is greater when the social relationships are closer, leading to the creation and modification of behaviors. For instance, Bruno et al. (2020) [30] investigated the impacts of political and psychological factors on recycling intentions using a causal model and found a correlation between public policy, environmental awareness, and recycling behavior. Rahman et al. (2022) [31] examined the significance of historical, cognitive, structural, and cultural embeddedness in metal recycling. Similarly, Reijonen et al. (2021) [32] identified a positive relationship between social norms and the sorting of plastic packaging. Building on these studies, this research acknowledges the influence of social pressure on recycling behavior, leading to the formulation of our second hypothesis:

Hypothesis 2 (H2).

Subjective norms have a positive effect on the intention to recycle cardboard boxes.

2.1.3. Perceived Behavioral Control

Perceived Behavioral Control (PBC) refers to the perceived ease or difficulty with which an individual can perform or control a specific behavior. When individuals have a more positive attitude toward a behavior, along with favorable subjective norms, a stronger sense of PBC is expected to emerge. This, in turn, leads to greater behavioral intention which has a higher likelihood of translating into actual behavior. For example, Ding et al. (2020) [33] found that perceived behavioral control significantly and positively affects residents’ acceptance of utilizing desalinated water. Similarly, Zhang et al. (2021) [34] investigated the impact of perceived behavioral control on residents’ intention to engage in waste classification. However, there is limited research exploring the relationship between perceived behavioral control and recycling behavior. Hence, our third hypothesis is as follows:

Hypothesis 3 (H3).

Perceived behavioral control regarding cardboard boxes positively influences recycling behavior.

2.1.4. Intention

Intention refers to an individual’s subjective assessment of the likelihood of engaging in a specific behavior, reflecting their willingness to perform the behavior. While there is a high correlation between behavioral intention and actual behavior, it is important to note that intention does not equate to behavior itself. For instance, Hao et al. (2019) [24] analyzed the influential factors affecting consumers’ willingness to pay for green packaging using a principal factor analysis. Similarly, Wang et al. (2019) [35] found a positive association between consumers’ intentions and their future behaviors. Based on the assumption that intention holds explanatory and predictive power over behavior, we propose our fourth hypothesis:

Hypothesis 4 (H4).

A stronger behavioral intention is expected to be associated with a higher likelihood of engaging in recycling behavior.

2.1.5. Benefits

Benefits refer to the material and psychological rewards that individuals seek to satisfy their desires. Rizzi et al. (2020) [36] investigated the relationship between consumers’ self-benefits and behaviors and discovered that when combined with effective communication and orientations, self-benefit appeals can encourage environmental behaviors. Similarly, Wang et al. (2019) [37] found a significant positive impact of economic motivation on willingness to engage in e-waste recycling behavior. However, the existing literature is limited with respect to the relationship between benefits and consumers’ willingness to recycle cardboard boxes. Therefore, our final hypothesis for this research study is:

Hypothesis 5 (H5).

There exists a positive relationship between recycling behavior and benefits.

Given the general recognition of the impact of benefits on consumers’ recycling behavior in the context of recycling cardboard boxes, we will incorporate benefits as influencing factors for an empirical analysis within the framework of the TPB model. The theoretical model of this paper is depicted in Figure 1.

2.2. The Post Station Service Platform Recycling Mechanism

The post station service platform serves as a third-party terminal logistics service system, providing last-mile delivery services to consumers. The platform consists of two key systems, as illustrated in Figure 2. The first system is the service system, which plays a crucial role in the forward logistics process. It allows consumers to temporarily store express parcels if they are unable to pick them up in a timely manner, thereby enhancing convenience for consumers. The second system is the recycling system, which serves as the starting point for the reverse logistics process and promotes resource utilization efficiency within express delivery companies. When consumers return packaging waste such as cardboard boxes to the post station service platform, the staff members screen and identify high-quality packaging materials to be reintroduced into the logistics transportation process. This integration of forward and reverse logistics functions within the express service post platform not only meets consumer demand for logistics services but also extends the life cycles of recyclable resources.

However, several issues persist in the implementation of this recycling mechanism, including a low consumer participation rate. In the following section, we will address these concerns and analyze the factors that influence consumers’ participation in the post station service platform recycling mechanism.

3. Methodology

3.1. Questionnaire and Scale Design

The content of the questionnaire in this study was designed to address the challenges faced by the post station service platform in the process of recycling cardboard boxes. It consisted of two parts: the first part collected respondents’ personal information, including gender, age, education background, monthly income, and frequency of online shopping and picking up express packages at the express service post. The second part collected relevant information about consumers’ recycling behaviors based on the theoretical model of this paper, with specific questions outlined in

Table 1. The structure of the questionnaire based on the actual recycling situation in China.

Variable	Item	Question Subject Matter
Attitude	AT1	Level of awareness regarding cardboard boxes’ recyclability
	AT2	Awareness of the usefulness of resource recycling
Subjective norms	SN1	The influence of family members
	SN2	The influence of friends
	SN3	The influence of the recycling policy generated by local government
	SN4	The influence of recycling activities holds by post station service platform
Perceived behavior control	PBC1	Knowledge about related policies or news about resource recycling
	PBC2	Knowledge about policy about sorting garbage
	PBC3	Knowledge about carbon peaking and carbon neutrality
	PBC4	Knowledge about activities generated by the post station service platform.
	PBC5	Knowledge about companies specializing in recycling activities
Benefits	BN1	The necessity of benefits provided by the post station service platform for consumers
	BN2	The price of recycling cardboard boxes for consumers
Intention	IN1	The degree of willingness to recycle cardboard boxes in the future
	IN2	The degree of willingness to call on people around you to recycle cardboard boxes in the future
	IN3	The relationship between recycling positivity and rewards from the post station service platform
	IN4	The relationship between recycling positivity and service level

. We employed measurement items AT1-AT2, SN1-SN4, PBC1-PBC5 and IN1-IN4 to describe the variables that influence recycling behavior, including attitude, subjective norms, perceived behavioral control, and intention, according to the theory of planned behavior (TPB). Additionally, we included questions related to benefits as shown in BN1-BN2 according to the extended theory of planned behavior (ETPB), which refer to the rewards provided by the post station service platform to consumers for their cooperation in recycling cardboard boxes.

The Likert scale is a measurement tool that captures respondents’ psychological states, and its ability to quantify a level of agreement or identification through their responses to questions can reflect the individuals’ comprehensive attitudes toward a behavior. Therefore, in this study, a five-point Likert scale was used to measure variables relating to recycling behavior. The scale followed the rule that the numbers 1–5 correspond to “totally disapprove or have no idea,” “disapprove or have no idea,” “neutral,” “approve or have knowledge,” and “strongly approve or strongly have knowledge,” respectively, in sequential order.

3.2. Data Collection

Considering the advantages of flexible and convenient data collection offered by web questionnaire software, we conducted our information collection using this method. Initially, a pre-survey was conducted to test the initial questionnaire and ensure its feasibility and accuracy. Subsequently, based on the results of reliability and validity analyses, items with substandard parameters were removed. Finally, a total of 700 questionnaires were collected using the final set of items.

The preliminary statistics of the reclaimed questionnaires, as shown in

Table 2. Descriptive statistics of individuals’ recycling behaviors and basic information.

Item	Selection	Frequency	Percent
Gender	Male	323	46.14%
	Female	377	53.86%
Age	20 or younger	45	6.43%
	21–30	264	37.71%
	31–40	176	25.14%
	41–50	137	19.57%
	51–60	64	9.14%
	Over 60	14	2.00%
Education	No degree	14	2.00%
	High school diploma	106	15.14%
	Bachelor’s degree	435	62.14%
	Master’s degree	128	18.29%
	Doctoral degree or greater	17	2.43%
Monthly income	1000 and lower	101	14.43%
	1000–5000	255	36.43%
	5000–10,000	206	29.43%
	10,000 and over	138	19.71%
Monthly frequency of shopping online	1–3	260	37.14%
	4–6	285	40.71%
	6 times or more	155	22.14%
Monthly frequency of picking up express at express service post	1–3	350	50.00%
	4–6	217	31.00%
	6 times or more	133	19.00%

, indicate that the sample consists of 46.14% men and 53.86% women, demonstrating a relatively balanced distribution of data. This implies that the perspectives on recycling behavior within the sample are representative of the majority, and the findings can be generalized to a larger population. Regarding the analysis of age attributes, it can be observed that the age groups 21–30 years, 31–40 years, and 41–50 years exhibit high response rates and are evenly distributed. This suggests that individuals across different age groups have notable interest in the issue of cardboard boxes recycling. The educational background of the survey participants can serve as an indicator of their understanding of the questionnaire. It is noteworthy that individuals with bachelor’s degree and greater comprise 82.86%. This implies that the questionnaire is reliable as it was comprehended by respondents with a higher level of education. In light of the cardboard box recycling issue, the basic information of the sample also includes variables such as monthly income, frequency of online shopping, and frequency of picking up express parcels at the express service post on a monthly basis. Specifically, the largest proportion of respondents falls within the monthly income range of 1000–5000 (36.43%), indicating their purchasing power. Furthermore, the respondents reported engaging in both online shopping and picking up express parcels at the service station on a monthly basis. This enhances the persuasiveness of the findings presented in this paper as it demonstrates their direct involvement and familiarity with the subject matter.

Furthermore, in order to examine consumers’ preferences for recycling cardboard boxes within the recycling mechanism, the questionnaire also included specific questions regarding the types of delivery packaging (as shown in Figure 3), the methods of disposing of cardboard boxes after picking up express parcels (as shown in Figure 4), and the reasons why consumers do not recycle cardboard boxes within the recycling mechanism (as shown in Figure 5).

As depicted in Figure 3, the questionnaire investigated five types of express packaging, namely, foam boxes, cardboard boxes, plastic bags, file pockets, and other types. The number of cardboard boxes, totaling 458, is the highest among all types of express packaging. Hence, researching the issue of recycling cardboard boxes holds practical significance and is feasible.

Subsequently, we examined how consumers manage packaging waste after collecting express parcels from the post station service platform, as illustrated in Figure 4. Regarding the recyclability of cardboard boxes, the questionnaire presented five disposal methods. Figure 4 reveals that half of the respondents choose to discard them directly in the trash, while only 17% of respondents are willing to place them in the recycling bin at the post station service platform. This finding highlights the presence of genuine challenges in the recycling mechanism implemented by express delivery companies, as evidenced by consumers’ behaviors in handling packaging waste.

Moreover, we conducted a preliminary analysis of the reasons why consumers do not recycle cardboard boxes using the questionnaire (as shown in Figure 5). It was found that 33% of respondents are unaware of which agencies can recycle cardboard boxes, indicating a need to strengthen the publicity efforts made with respect to the recycling mechanism employed by express delivery companies. Additionally, a quarter of the respondents have limited knowledge about recycling and are unaware that cardboard boxes can be recycled. Furthermore, 24% of respondents believe that recycling cardboard boxes is time-consuming, suggesting that many consumers do not yet have a well-established awareness of recycling.

3.3. Model

With the aim of exploring the factors influencing consumer recycling behavior, we conducted research on the determinants of recycling behavior concerning cardboard boxes among consumers. To predict the likelihood of such behavior, we employed logistic regression, which is a generalized linear model. There are three common types of logistic regression, each suited to different characteristics of the dependent variables, such as number and nature. Considering behavioral characteristics with respect to recycling cardboard boxes and the specific context of our study, we chose to utilize multinomial logistic regression as the analytical tool. This type of regression is appropriate when the dependent variables have more than two categories and are logically related without a specific order. By employing this approach, we aimed to provide a more robust and scientifically grounded basis for policy recommendations.

Multinomial logistic regression, as a supervised classification machine learning algorithm, extends logistic regression [38,39]. If the dependent variable has J categories, then the multinomial logistic model will have J-1 logits and the last category (i.e., the Jth category) is used as the reference category. The formula is presented in detail in Equation (1), where P(.) refers to the probability of occurrence of the corresponding category, ${\beta }_{ji}$ is the coefficient of regression for the model, and $x_i$ indicates the explanatory variables which may influence the explained variable.

$$\begin{gathered} \mathit{ln}\left(\frac{P\left(y=j\right|x)}{P\left(y=J\right|x)}\right)={\alpha }_j+\sum _{i=1}^K{\beta }_{ji}{x}_i \\ i=\mathrm{1,2},···,k;j=\mathrm{1,2},···,J-1 \end{gathered}$$

(1)

In combination with the actual issue of recycling cardboard boxes, the explanation of Equation (1) is as follows: j denotes consumer methods of disposing of packaging waste, namely, j = taking reluctant, neutral, and positive action on attending the recycling mechanism provided by the express delivery companies, respectively. We then regard the willingness to participate in recycling activity as a reference item; thus, there are two multinomial logistic regression models in our research. The regression model affecting the ratio of the probability of consumers choosing to not participate recycling activities is shown in Equation (2), and the regression model affecting the ratio of the probability of adopting a neutral attitude toward recycling activities is shown in Equation (3).

$$\mathit{ln}\left(\frac{P_{unwilling}}{P_{willing}}\right)={\alpha }_{unwilling}+{\sum }_{i=1}^K{\beta }_{unwilling\hspace{1em}i}{x}_i$$

(2)

$$\mathit{ln}\left(\frac{P_{neutual}}{P_{willing}}\right)={\alpha }_{neutual}+{\sum }_{i=1}^K{\beta }_{neutual\hspace{1em}i}{x}_i$$

(3)

Since the dependent variable of logistic regression is not a continuous variable, we used the logit (i.e., logarithmic occurrence ratio: ln[p/(1 − p)]). To explain the logistic regression coefficient, the natural exponents are taken on both sides of the model equation. The odds ratio (OR) was then used to explain the effects of the independent variables on the overall probability ratio, and its expression is shown in Equation (4).

$$\left(\frac{P\left(y=j\right|x)}{P\left(y=J\right|x)}\right)=e^{{\alpha }_j+\sum _{i=1}^K{\beta }_{ji}{x}_i}$$

(4)

3.4. Data Analysis Method

In our research, we conducted a thorough analysis of the questionnaire data using various methods and models, as visually presented in Figure 6. Firstly, we performed a reliability analysis using SPSS 24.0 to assess the level of trustworthiness of the data. Additionally, we conducted a validity analysis to evaluate the appropriateness of the questionnaire design. Secondly, we employed a principal component analysis (PCA) to extract common factors from the variables that influence recycling behavior. Finally, we predicted recycling behavior using multinomial logistic regression.

4. Analysis

4.1. Reliability and Validity Analysis

During the investigation process, various factors can affect the quality of the questionnaire results, including the respondents’ individual reasons. Therefore, it is necessary to assess the reliability and consistency of the questionnaire before conducting the analysis to ensure the accuracy of the results. A reliability analysis is a measurement tool used to assess the credibility, consistency, and stability of a scale and serves as the initial step in analyzing questionnaire data. In this study, we employed Cronbach’s alpha coefficient and used SPSS 24.0 to examine the reliability of the 700 questionnaires. The overall Cronbach’s alpha coefficient for the questionnaire was found to be 0.902 (>0.70), indicating a high level of reliability. This suggests that the collected data are suitable for further analysis [40].

In order to simplify the data analysis process, we opted to employ an exploratory factor analysis (EFA) to reduce the number and dimensions of variables obtained from the questionnaire. Before extracting the common factors using the EFA, it is essential to conduct a validity analysis to ensure the appropriateness of the item design. The validity analysis examines the soundness and relevance of the questionnaire items.

To assess the questionnaire’s ability to accurately measure the intended variables, it is crucial to conduct a validity analysis, which examines the data distribution and the level of correlation between the questions. The results of the validity analysis are presented in

Table 3. The results of the KMO and Bartlett’s test.

Kaiser-Meyer-Olkin measure of sampling adequacy		0.884
Bartlett’s test of sphericity	Approximate chi-square	7210.404
	Df	136
	Sig.	0.000

. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy is 0.884 (>0.7), with a significance value of 0.000 (<0), indicating a strong correlation among the variables and demonstrating that the collected data are suitable for an exploratory factor analysis.

4.2. Factor Analysis

To address the complexity of the problem, irrelevant variables are eliminated by examining the correlation coefficient matrix between variables. After conducting the Kaiser-Meyer-Olkin and Bartlett tests, this section proceeds with extracting common factors from the 17 items in the questionnaire using a principal component analysis (PCA). A PCA is a statistical method that reduces dimensionality by transforming a set of n indicators into m (m < n) indicators while minimizing information loss. The transformed composite indicators become the co-factors, each representing a linear combination of the original variables.

The total variances explaining the variables are presented in the Supporting Information file. Five factors, each with an eigenvalue greater than 1, are identified as the common factors. These five factors account for a cumulative variance contribution rate of 76.073%, indicating that the model provides a better description of the observed data while minimizing information loss. This finding suggests that the model, based on the five co-factors, aligns well with the theoretical framework of this paper. Consequently, the results obtained from this survey hold valuable insights.

Once the number of co-factors has been determined, the next step is to examine the relationships between the variables and the co-factors. For this purpose, we employ the maximum variance method for rotation in the subsequent analysis. Subsequently, all variables are categorized based on factor loadings greater than 0.6. Factor loadings provide a measure of the associations between variables and common factors. A higher absolute value of the factor loading indicates a stronger relationship between the variable and the co-factor, suggesting that the factor is a better representation of the variable. The rotation component matrix, which demonstrates the relationships between variables and common factors, is calculated and is presented in

Table 4. The rotation component matrix.

	Common Factor
	1	2	3	4	5
PBC1	0.801
PBC5	0.795
PBC4	0.791
PBC3	0.755
PBC2	0.750
SN1		0.850
SN2		0.835
SN4		0.813
SN3		0.803
IN1			0.860
IN3			0.808
IN4			0.800
IN2			0.793
BN2				0.848
BN1				0.843
AT2					0.869
AT1					0.794

Extracting method: principal component analysis; Rotating method: varimax with Kaiser normalization.

.

The first common factor exhibits relatively higher loadings for five variables: PBC1 (0.801), PBC5 (0.795), PBC4 (0.791), PBC3 (0.755), and PBC2 (0.750). These variables pertain to the perceived difficulty of engaging in a certain behavior. Therefore, the first principal component is renamed PBC (perceived behavioral control). The second common factor demonstrates high factor loadings for the following variables: SN1 (0.850), SN2 (0.835), SN4 (0.813), and SN3 (0.803), which indicate the influence of others’ evaluations on decision making. Thus, the second principal component is referred to as SN (subjective norms). Among the loadings of the third common factor, four variables exhibit higher loadings: IN1 (0.860), IN3 (0.808), IN4 (0.800), and IN2 (0.793), reflecting the intention behind a behavior. Consequently, the third principal component is denoted as IN (intention). The fourth common factor displays higher loadings for two variables: BN2 (0.848) and BN1 (0.843), which represent the benefits associated with a particular behavior. Thus, the fourth principal component is labeled BN (benefits). Lastly, the fifth common factor demonstrates higher loadings for AT2 (0.869) and AT1 (0.794), indicating the attitude toward a behavior. Therefore, the fifth principal component is designated as AT (attitude).

The component score coefficient matrix illustrates the relationship between each indicator variable and the extracted common factors. A higher coefficient score for a specific common factor indicates a stronger association between the indicator variable and that particular common factor. The factor score function for each common factor is presented in

Table 5. Component score coefficient matrix.

Item	Component
	1	2	3	4	5
AT1	−0.014	−0.049	−0.107	0.020	0.578
AT2	−0.058	−0.061	−0.024	−0.093	0.654
SN1	−0.077	0.353	−0.023	−0.104	0.003
SN2	−0.059	0.351	−0.030	−0.070	−0.087
SN3	−0.066	0.321	−0.020	−0.068	−0.006
SN4	−0.083	0.345	−0.018	−0.062	−0.063
PBC1	0.283	−0.038	−0.028	−0.041	−0.070
PBC2	0.321	−0.099	0.050	−0.212	0.041
PBC3	0.278	−0.086	0.014	−0.090	0.048
PBC4	0.266	−0.051	−0.039	0.013	−0.050
PBC5	0.277	−0.045	−0.035	0.002	−0.095
BN1	−0.094	−0.098	0.005	0.625	−0.030
BN2	−0.089	−0.073	−0.030	0.623	−0.047
IN1	0.011	−0.016	0.321	−0.042	−0.056
IN2	0.050	−0.041	0.304	−0.042	−0.086
IN3	−0.043	−0.034	0.304	−0.003	−0.011
IN4	−0.045	−0.026	0.313	0.024	−0.088

Extraction method: principal component analysis; Rotation method: varimax with Kaiser normalization.

.

PBC = −0.014 AT1 − 0.058 AT2 − 0.077 SN1 − 0.059 SN2 − 0.066 SN3 − 0.083 SN4 + 0.283 PBC1 + 0.321 PBC2 + 0.278 PBC3 + 0.266 PBC4 + 0.277 PBC5 − 0.094 BN1 − 0.089 BN2 + 0.011 IN1+0.050 IN2 − 0.043 IN3 − 0.045 IN4

(5)

SN = −0.049 AT1 − 0.061 AT2 + 0.353 SN1 + 0.351 SN2 + 0.321 SN3 + 0.345 SN4 − 0.038 PBC1 − 0.099 PBC2 − 0.086 PBC3 − 0.051 PBC4 − 0.045 PBC5 − 0.098 BN1 − 0.073 BN2 − 0.016 IN1 − 0.041 IN2 − 0.034 IN3 − 0.026 IN4

(6)

IN = −0.107 AT1 − 0.024 AT2 − 0.023 SN1 − 0.030 SN2 − 0.020 SN3 − 0.018 SN4 − 0.028 PBC1 + 0.050 PBC2 + 0.014 PBC3 − 0.039 PBC4 − 0.035 PBC5 + 0.005 BN1 − 0.030 BN2 + 0.321 IN1 + 0.304 IN2 + 0.304 IN3 + 0.313 IN4

(7)

BN = 0.020 AT1 − 0.093 AT2 − 0.104 SN1 − 0.070 SN2 − 0.068 SN3 − 0.062 SN4 − 0.041 PBC1 − 0.212 PBC2 − 0.090 PBC3 + 0.013 PBC4 + 0.002 PBC5 + 0.625 BN1 + 0.623 BN2 − 0.042 IN1 − 0.042 IN2 − 0.003 IN3 + 0.024 IN4

(8)

AT = 0.578 AT1 + 0.654 AT2 + 0.003 SN1 − 0.087 SN2 − 0.006 SN3 − 0.063 SN4 − 0.070 PBC1 + 0.041 PBC2 + 0.048 PBC3 − 0.050 PBC4 − 0.095 PBC5 − 0.030 BN1 − 0.047 BN2 − 0.056 IN1 − 0.086 IN2 − 0.011 IN3 − 0.088 IN4

(9)

From Equations (5)–(9), the variables PBC, SN, IN, BN, and AT can be utilized as substitutes for all the original variables. These variables effectively capture and assess consumer inclination to deposit cardboard boxes into the recycling bins offered by express post station service stations.

In the subsequent section, these extracted co-factors are treated as independent variables in the regression analysis to examine the relationship between recycling behavior and these factors.

4.3. Regression Analysis

This study examines the willingness of consumers to participate in the recycling activities proposed by express post station service stations. In this section, the frequency of dropping cardboard boxes into the recycling bin is used as the dependent variable in the multinomial logistic regression analysis. The co-factors extracted in the previous section are employed as independent variables in the analysis. Two regression models are developed based on different behaviors, and the results of the regression analysis, presented in

Table 6. The result of parameter estimates.

Selection of Recycling Behaviors	Variables	B	Std. Error	Wald	Df	Sig.	Exp(B)	95% Confidence Interval for Exp(B)
								Lower Bound	Upper Bound
Unwilling to participate in recycling behavior	Intercept	−0.161	0.153	1.100	1	0.294
	AT	−1.626	0.159	104.520	1	0.000	0.197	0.144	0.269
	SN	−1.209	0.142	72.781	1	0.000	0.299	0.226	0.394
	PBC	−0.211	0.130	2.617	1	0.106	0.810	0.627	1.046
	BN	−0.827	0.135	37.622	1	0.000	0.437	0.336	0.570
	IN	−0.274	0.128	4.605	1	0.032	0.760	0.592	0.977
Neutral stance toward recycling behavior	Intercept	0.901	0.116	60.715	1	0.000
	AT	−0.866	0.107	66.084	1	0.000	0.421	0.341	0.518
	SN	−0.673	0.105	40.797	1	0.000	0.510	0.415	0.627
	PBC	−0.224	0.107	4.406	1	0.036	0.799	0.648	0.985
	BN	−0.444	0.099	20.285	1	0.000	0.641	0.529	0.778
	IN	−0.194	0.103	3.514	1	0.061	0.824	0.673	1.009

The reference category is: willing to participate in recycling behavior.

, illustrate the extent of the influence exerted by the various independent variables and predict the likelihood of recycling behavior occurring.

In this regression model, there are three types of recycling behavior: (a) unwillingness to participate in the recycling activity, (b) neutral stance toward recycling behavior, and (c) willingness to participate in the recycling activity. The third behavior is used as the reference category for the first two behaviors. The probabilities of the three behaviors are denoted as $P_{unwilling}$,$P_{neutual},$and $P_{willing},$ respectively.

4.3.1. Model 1

The results of the first model are presented in Equation (10), where we set the behavior of being willing to participate in recycling activity as the reference category. When comparing the willingness and reluctance to participate in recycling activities provided by the express post station service station, the variables AT, SN, BN, and IN show statistical significance with p-values less than 0.05.

$$\mathit{ln}\left(\frac{P_{unwilling}}{P_{willing}}\right)=-0.161-1.626AT-1.209SN-0.827BN-0.274IN$$

(10)

(i)

The regression coefficient for AT is −1.626, suggesting that individuals with a stronger recycling attitude are more likely to choose willingness to recycle rather than unwillingness. Furthermore, the odds ratio (OR) for AT is 0.197, indicating that the probability of being unwilling to participate in the recycling activity is 0.197 times that of being willing to participate. In other words, the likelihood of choosing willingness is 5.076 times higher than choosing unwillingness.

(ii)

The regression coefficient for SN is −1.209, indicating that individuals who are more susceptible to outside influences have a higher probability of choosing recycling behavior instead of unwillingness. The odds ratio (OR) for SN is 0.299, suggesting that the likelihood of being unwilling to participate in the recycling activity is 0.299 times that of choosing a recycling behavior with a positive attitude. In other words, the probability of choosing willingness is 3.344 times higher than choosing unwillingness.

(iii)

The regression coefficient for BN is −0.827, suggesting that individuals who place importance on personal benefits in their daily lives are more likely to choose recycling behavior. The odds ratio (OR) for BN is 0.437, indicating that the likelihood of being unwilling to participate in the recycling activity is 0.437 times that of choosing willingness. In other words, the probability of choosing willingness is 2.288 times higher than the probability of choosing unwillingness.

(iv)

The regression coefficient for IN is −0.274, suggesting that individuals with a stronger intention to act are more likely to choose recycling behavior rather than being unwilling. Furthermore, the odds ratio (OR) for IN is 0.760, indicating that the probability of being unwilling is 0.760 times that of choosing willingness. In other words, the likelihood of choosing willingness is 1.316 times higher than the likelihood of choosing unwillingness.

The interpretation of the coefficient results from the logistic regression models indicates that the factors can be ranked in order of importance as follows: AT, SN, BN, and IN. Among these factors, changes in AT have the highest probability of causing a change in behavior. This suggests that individuals’ attitudes toward environmental protection have the greatest impact on behavioral choices and changes. Changes in SN and BN also influence recycling behavior, indicating that evaluations of others and the perceived benefits from participating in recycling activities have significant influences on behavior choices. On the other hand, changes in IN have the smallest probability of bringing about a change in recycling behavior, indicating that the desire or intention to engage in a certain behavior has the least effect on the decision-making process or behavior change. Therefore, Hypotheses H1, H2, H4, and H5 are supported, while H3 is rejected.

4.3.2. Model 2

In the second model, as shown in Equation (11), the reference category is still the behavior of being willing to participate in the recycling activity. When comparing recycling behavior with a neutral attitude and a willing attitude, the common factors AT, SN, BN, and PBC are statistically significant, as indicated by p-values of less than 0.05.

$$\mathit{ln}\left(\frac{P_{neutual}}{P_{willing}}\right)=0.901-0.866AT-0.673SN-0.224PBC-0.444BN$$

(11)

(i)

The regression coefficient for AT is −0.866, suggesting that individuals with stronger recycling attitudes are more likely to adopt a willing attitude toward recycling rather than being neutral. Furthermore, the odds ratio (OR) for AT is 0.421, indicating that the probability of having a neutral attitude toward recycling behavior is 0.421 times lower compared to having a willing attitude. In other words, individuals are 2.375 times more likely to adopt a willing attitude than a neutral attitude in their behavioral choices.

(ii)

The regression coefficient for SN is −0.673, suggesting that individuals who are more susceptible to outside influences are more likely to choose recycling behavior rather than being neutral. Additionally, the odds ratio (OR) for SN is 0.510, indicating that individuals are 0.510 times more likely to adopt a neutral attitude toward recycling behavior compared to a willing attitude. Alternatively, it can be interpreted as the probability of choosing a willing attitude toward recycling behavior being 1.960 times higher than choosing a neutral attitude.

(iii)

The regression coefficient for BN is −0.444, suggesting that as the rewards of recycling increase, individuals are more likely to adopt a willing attitude toward recycling. The odds ratio (OR) for BN is 0.641, indicating that compared to choosing a willing attitude toward recycling, the probability of choosing a neutral attitude is 0.641 times lower. In other words, the probability of choosing a willing attitude toward recycling is 1.560 times higher than the probability of choosing a neutral attitude.

(iv)

The regression coefficient for PBC is −0.224, suggesting that as the ease of recycling behavior increases, individuals are more likely to adopt a willing attitude toward recycling rather than a neutral attitude. The odds ratio (OR) for PBC is 0.799, indicating that compared to choosing a willing attitude toward recycling, the probability of choosing a neutral attitude is 0.799 times lower. In other words, the probability of choosing a willing attitude toward recycling is 1.252 times higher than the probability of choosing a neutral attitude.

The interpretation of the coefficient results from the logistic regression models reveals that the factors’ importance ranks in the order of AT, SN, BN, and PBC when comparing the choice between being willing to recycle and adopting a neutral attitude toward recycling. Specifically, a change in AT has the highest probability of causing a change in behavior, indicating that recycling attitude has the greatest impact on behavioral change. The probabilities associated with the common factors SN, BN, and PBC in influencing a change in behavior are relatively similar, suggesting that the evaluation from others, recycling benefits, and the ease of recycling behavior have comparable influences. Therefore, Hypotheses H1, H2, H3, and H5 are supported, while H4 is rejected.

5. Conclusions

The issue of resource consumption and environmental pollution has emerged as a global problem, capturing increased attention from governments and industries worldwide. China, as a country facing significant challenges in the conservation of resources, has embarked on extensive and comprehensive initiatives to promote and support the recycling industry in recent years.

Guided by the extended theory of planned behavior and in response to practical issues encountered in the operation of the recycling system implemented by express post station service stations, this study undertook the design and collection of 700 questionnaires from diverse individuals in China. Utilizing a principal component analysis, the study extracted five factors, namely, AT, SN, PBC, BN, and IN, from the data collected from the 700 questionnaires. These five factors served as independent variables in a multinomial logit model, enabling an analysis of their impacts on consumer participation in the activity of recycling cardboard boxes, which is facilitated by the station. The findings from this analysis will inform future research endeavors.

Based on the preceding analysis, the following conclusions can be drawn:

(1)

The study reveals several key findings. First, there is a need to enhance consumers’ recycling attitudes. While 7% of respondents repurpose cardboard boxes at home and 24% sell them to ragmen, a significant proportion (51%) still dispose of recyclable cardboard boxes in the regular trash. Furthermore, 24% of respondents consider recycling cardboard boxes to be a waste of time. Additionally, the study highlights a lack of knowledge about recycling, with 20% of respondents unaware that cardboard boxes can be recycled. To address these issues, the government should prioritize efforts and allocate resources to influence consumer attitudes toward recycling cardboard boxes, as this is deemed the most effective way to improve the current situation. Additionally, it is important for express delivery companies to address shortcomings in their recycling mechanisms, particularly in terms of inadequate publicity and awareness campaigns.

(2)

The questionnaire designed via the ETPB identified 17 factors that influence consumer recycling behavior (see Table 1). Through a factor analysis, five principal factors were extracted: AT, SN, PBC, IN, and BN. AT represents individuals’ positive or negative attitudes toward recycling behavior. SN captures the influence of external pressures, including family members, social connections, express delivery companies, and the government, on consumers’ recycling behaviors. PBC reflects the perceived control individuals have over their recycling behavior. IN signifies the subjective determination of the likelihood that consumers will engage in recycling actions. Finally, BN pertains to the direct or indirect benefits associated with consumer participation in the recycling mechanism. After conducting a multinomial logistic regression analysis, the ranking of common factors influencing the consumers’ recycling behaviors is as follows. When comparing willingness or reluctance to participate in the recycling mechanism provided by express delivery companies, four common factors (attitude, subjective norms, benefits, and intention) have a significant impact on recycling behavior. Attitude has the highest influence, causing the likelihood of choosing willingness compared to unwillingness to be 5.076 times greater. The influence of subjective norms causes the likelihood of choosing willingness to be 3.344 times greater. The influence of benefits causes the likelihood of choosing willingness to be 2.288 times greater, and the influence of intention causes the likelihood of choosing willingness to be 1.316 times greater. In the context of choosing willingness or neutral action in the recycling mechanism, four common factors (attitude, subjective norms, benefits, and perceived behavioral control) impact behavior. Attitude has a likelihood 2.375 times greater for choosing willingness compared to choosing neutral action. The influence of subjective norms causes the likelihood of choosing willingness to be 1.960 times greater, the influence of benefits causes the likelihood to be 1.560 times greater, and the influence of perceived behavioral control causes the likelihood to be 1.252 times greater. Based on these findings, governments and express delivery companies can focus on changing consumers’ attitudes toward recycling as the most effective means of improving the current situation. By addressing and improving consumers’ attitudes, positive behavioral change toward recycling can be encouraged.

5.1. Practical Implications

The conclusions drawn from the combination of the PCA and multinomial logistic regression analysis offer valuable insights for relevant departments and scholars. By examining the factors that influence recycling behavior, we can gain a deeper understanding of the underlying dynamics and make informed decisions and collaborative efforts accordingly. These findings have practical implications for both government and business sectors, allowing them to develop targeted strategies and interventions to promote and improve recycling behavior.

There are two measures that the government can implement to increase the recycling rate. Firstly, the local department should adjust the focus of advocacy content. The questionnaire data reveal that consumers in China have a basic awareness of recycling and understand its positive impact on the environment. However, there is a lack of knowledge regarding the specific recycling practices and processes. Therefore, government communication efforts should emphasize which items can be recycled and provide information about the recycling process. Doing so will help consumers achieve have a better understanding of the value of recyclable materials and their potential economic benefits. Such interventions are more effective at changing individuals’ environmental concerns and promoting their participation in various recycling activities organized by relevant departments when compared to general environmental advocacy. Secondly, sustainable development, as an eco-economic model, requires financial investment. Therefore, the government should provide financial support to companies engaged in recycling efforts, ensuring that they have sufficient resources for sustainable development. This financial assistance will create more opportunities for recycling businesses to thrive and expand their operations, ultimately contributing to a higher recycling rate and a more sustainable society.

There are three practical recommendations for express postal service stations. Firstly, improving the quality of promotional work is crucial in addressing the low recycling rates. The questionnaire data highlight that some consumers are completely unaware of recycling activities, indicating gaps in the post station service platform’ publicity work. Therefore, after each policy or initiative is introduced, station staff should employ various methods, such as displaying posters at the station or engaging in direct communication with consumers, to raise awareness about the recycling activities. Secondly, post station service platform should optimize their incentives to encourage consumer participation in the recycling process. The regression model findings indicate that benefits have a positive influence on consumers’ willingness to participate in recycling activities. Therefore, implementing a reasonable incentive mechanism can effectively enhance consumer engagement in recycling. This can be achieved by providing rewards, discounts, or other incentives to individuals who actively participate in recycling efforts. Thirdly, the post should prioritize innovation in service quality. As a service provider, the post should not only ensure the quality of its core services but also explore new services to enhance consumer satisfaction. In the context of recycling, the post can offer services such as telephone booking or door-to-door collection to reduce the logistical challenges faced by consumers in attending recycling activities. By continuously improving service quality and convenience, the post can further encourage consumers to participate in recycling initiatives.

5.2. Limitation and Further Research

While this study endeavored to construct a model that closely mimics real-life scenarios to investigate the factors influencing recycling behavior, certain objective constraints still leave room for further research. Firstly, there is a limitation in the sample collection process, as the consumer groups gathered for this research are primarily from the less-developed northwest region of China. Given the regional disparity in resource recycling development across China, the analysis results may be influenced by regional differences. Moreover, this study introduced the extended theory of planned behavior, (ETPB) based on the theory of planned behavior (TPB) and practical contexts, aiming to capture real-world dynamics as comprehensively as possible. However, the TPB is an open theoretical framework, and there may be other factors beyond those proposed in this paper that can also influence recycling behavior. Hence, future research could explore additional variables and their potential impacts on recycling behavior to gain a more comprehensive understanding of the phenomenon.

As China’s courier industry continues to grow, the use of recyclable materials for courier packaging is expected to increase. Consequently, the recycling of cardboard boxes remains a topic of interest across various industries. Several potential areas of future research can be considered in this regard. Firstly, investigating the influences of diverse economic conditions and development levels on recycling behavior in different regions would be valuable. Regional disparities may impact the effectiveness of recycling initiatives and understanding these dynamics can inform targeted interventions. Secondly, since cardboard box recycling is still in its developmental phase, it may be worthwhile to identify new factors that influence consumer recycling behavior, aligning with the evolving nature of recycling activities. Building an improved evaluation index system for assessing recycling willingness can contribute to a more comprehensive understanding of recycling behavior and inform effective strategies.