4.1. Basic Information of Respondents
In the sample of rural residents participating in the garbage exchange supermarket exchange (see
Table 1), the proportion of women is higher than that of men. Most rural residents are over 60 years old, and the lowest proportion of rural residents is between 20–29 years old. Among the residents, the educational level is mainly primary school and below; the annual income is mainly 5000 yuan and below, with those earning 10,000–15,000 comprising the smallest group. Farming is the main occupation and most rural residents not only engage in agriculture but also go out to work during quiet farming seasons. Almost all respondents were locals, accounting for 99.1% of the sample, and very few were temporary residents.
The results of the survey found that almost all residents participate in garbage exchange activities (99.7%), and the coverage rate of villagers is very high. Most of the exchanged garbage comprises daily waste (see
Table 2), of which plastic bags are the highest category (85.5%), followed by plastic drinking bottles, cigarette shells, and cigarette butts. For exchange items, residents mainly choose daily necessities, among which salt is the highest (62.3%), followed by detergent (55.6%). Most residents (52.5%) visit garbage exchange supermarkets twice a month, or once a week (16.7%) for those who generate more household waste. Most areas have not implemented a points management system, so the participation rate in those schemes is not high. In the sample, only 24.7% of residents have participation-related points. When asked how they learned about the exchange supermarket, 56.5% were informed by friends and neighbors, 47.8% were informed by “announcements posted in the village”, and 19.1% were informed about the supermarket through the distribution of “a letter to residents”. These results show that the spread of information in rural areas is mainly by word of mouth. The garbage exchange supermarket also held related activities, such as agricultural product exhibitions and sales. Only 5.9% of the residents said they participated in such related activities. For ecological compensation, most of the residents are unaware of the concept and only 10.3% have heard of, and have a certain understanding of, ecological compensation.
4.3. Confirmatory Factor Analysis
This study carries out confirmatory factor analysis on the relationship between the latent variables and their measurements. Before the factor analysis, the Cronbach’s α value and combined reliability of each variable are first tested for reliability of the scale. The results show that the Cronbach’s α value is 0.802, which was greater than 0.7, indicating that the measurement items have good correlation. The Bartlett sphericity test also shows a significant value of 0.000, which is less than 0.05, indicating that the data have good correlation and are suitable for factor analysis. Second, the indicators of each dimension are analyzed by SPSS and the Cronbach’s α value of each dimension variable is obtained. From the perspective of validity, this study has a certain guarantee. From the perspective of content validity, a certain guarantee is obtained with the help of maturity scale, expert opinions, and group discussion, plus the verification of the results of the pre-survey. Finally, confirmatory factor analysis is carried out with Amos software to obtain the standardized factor loadings of each measurement variable. The combined reliability (CR) and average variance extraction value (AVE) of each dimension are calculated using the formula and Excel tool, where a CR value greater than 0.7 indicates high consistency, and an AVE value greater than 0.5 indicates that each dimension has discriminant validity.
In this study, perception is a variable with a multi-dimensional structure. Eight two-dimensional (2D) indicators were selected to measure the perception dimension. Given that this dimension is 2D, each first level needs to be evaluated, and the two sub-levels can be jointly validated for factor analysis.
Table 5 shows the results. The Cronbach’s α values of benefit and cost perceptions are 0.810 and 0.861, respectively. The CR values are 0.8096 and 0.8634, respectively, and each variable is greater than 0.7, indicating that the scale has a good internal consistency. Thus, the model has good reliability. The standardized loading values of each variable range from 0.52 to 0.85, which are all greater than 0.5, indicating that each variable can explain the perception variable. Moreover, the AVE values of the perceptual variables are 0.4659 and 0.6783, respectively, which are greater than 0.5 and proves that this dimension has good validity.
In this study, satisfaction is a one-dimensional structural variable. This study selects five measurement indicators to examine this dimension and builds a corresponding measurement model, as shown in
Figure 3. Analysis using Amos and SPSS software shows that the Cronbach’s α of satisfaction is 0.831 and the CR is 0.8318, both of which are greater than 0.7, indicating that the reliability of this dimension is high, and the model reliability is good. The standardized loading values of the four measurement variables range from 0.69 to 0.79, reaching a significant level, indicating that each measurement item can explain the satisfaction variable. Moreover, the AVE of this variable is 0.5534, which is greater than 0.5, indicating that this dimension has good validity.
In this study, attitude is a one-dimensional variable, and is explained using four variables. Factor analysis is carried out using SPSS, and an attitude model is constructed by using Amos. The results show that the Cronbach’s α value of attitude is 0.791, which is greater than 0.7, indicating that this dimension has a relatively high value and good consistency. Then, exploratory factor analysis is carried out and shows that the standardized loading value of each factor range from 0.52 to 0.89, which is greater than 0.5, and the CR value is 0.7889; thus, the measurement has good validity. The AVE value is 0.4963, which is close to 0.5, and indicates acceptable reliability.
The dimension of environmentally responsible behavior is two-dimensional and two factors are extracted and tested separately. The overall Cronbach’s alpha value of environmentally responsible behavior is 0.854, which is greater than 0.7, indicating that the scale has good internal consistency. The factor loadings of each variable range are 0.670–0.880, which are greater than 0.500, indicating that the scale has good internal consistency and validity. Next, the standardized loadings of the two dimensions of the compliance-type and the promotion-type environmentally responsible behaviors are calculated. The AVE values are 0.6154 and 0.5665, which are greater than 0.500, and the CR values are 0.8263 and 0.8391, which are higher than 0.7. Thus, the dimension of environmentally responsible behavior has good reliability and validity.
4.4. Structural Model Checking and Correction
This study has five latent variables in the measurement model, including the exogenous latent variables of cost and benefit perceptions and the endogenous latent variables of satisfaction, attitude, and environmentally responsible behavior. The measurement items are five for benefit perception, three for cost perception, five for satisfaction, four for attitude, and seven for environmentally responsible behavior.
Figure 3 shows the structural equation model diagram.
The initial measurement model is calculated by using Amos, and
Figure 4 shows the model fitting result. The fitness of the overall model shows that the chi-square value is 2.668, which is less than 3, and RMSEA = 0.072, which is less than 0.08; thus, good fitness is reached. However, in the initial model results, GFI = 0.850, AGFI = 0.814, CFI = 0.880, NNFI = 0.863, and IFI = 0.881, all are less than 0.9 and none reach the recommended value standard. Thus, the initial model needs to be adjusted and optimized. The initial model is corrected with reference to the correction index to further improve the overall accuracy. The addition of covariation relationships between e30 and e32, e28 and e29, and e30 and e31 shows the gradual improvement for each index. The modified model operation results show (see
Table 6) that except for the values of AGFI and NFI which are lower than 0.9, the rest of the values are within the range of valid values, and the model fit is generally good and acceptable.
4.5. Structural Equation Model Analysis
The model path analysis is further carried out on the modified model.
Table 7 shows the results.
The results of the revised model show that residents’ perceptions of the benefits of garbage exchange supermarkets have a positive effect on attitudes (p < 0.05), and the path critical value (CR) is 2.00, that is, H1 is supported. Residents participate in garbage exchange and perceive the benefits they obtain. Through analysis, rural residents believe that participating in garbage exchange can obtain certain social benefits (rural image) and environmental benefits (water and air improvement). In the benefit perception, residents have the deepest understanding of environmental benefit, and the explanatory value of reduction of waste in village is the largest. When asked about their attitude towards the garbage exchange supermarket during the field investigation, the residents said, “Since the establishment of the supermarket, basically no garbage can be seen in the village, and the environment has improved a lot. We think this supermarket is very well built, which is for the consideration of our people.” At the same time, the relationship between the residents’ perceptions of the benefits of garbage exchange supermarkets and residents’ satisfaction is proven, p < 0.001, the path critical value (CR) is 5.660, and thus H4 is established. In terms of satisfaction with the services provided by the supermarket, the residents said, “We are very satisfied, the staff are all locals, and our daily garbage is directly exchanged on the exchange day, and the salt, soap and dishwashing liquid exchanged back are very useful, so there is basically no need to spend money to buy daily necessities”.
Resident satisfaction and attitude towards garbage exchange supermarkets are also confirmed. The path coefficient is 4.439, p < 0.001, reaching saturation and significance, indicating that residents’ satisfaction with garbage exchange supermarkets significantly affects residents’ attitudes. Thus, H6 is established. During the interview, residents also stated that the service personnel of the supermarket are all locals and the exchange items provided are household goods. The weekly exchange time is highly appropriate and the exchange places are in the central areas, which is very convenient. In general, residents have a positive attitude towards supermarkets and are also satisfied with the services, exchanges, and management models provided. This satisfaction can promote the residents’ support for the garbage exchange supermarket to a certain extent, which also reflects the residents’ attitudes towards this initiative.
Second, the relationship between residents’ cost perceptions and attitudes toward garbage exchange supermarkets is not verified. The path critical ratio (CR) value of cost perception to support attitude is −2.153 and the p value is 0.312, which is greater than 0.05. Therefore, residents’ perceptions of the cost of participating in garbage exchange is unrelated to residents’ attitudes; therefore, H2 does not hold. Cost and benefit perceptions are the manifestations of different perception dimensions. According to social exchange theory, when residents engage in an activity, the trade-off between perceived costs and perceived benefits determines their participation. In the context of garbage exchange supermarkets, residents believe that the cost of participating in the exchange can be ignored. At the same time, the on-site investigation reveals that, due to the inconsistent distance from the exchange point, residents who are relatively close to the supermarket generally do not need to queue and wait for those who are farther away to complete the exchange; however, the residents who are a little further away go earlier and queue in front of the supermarket. In this dimension, residents’ perceptions vary greatly, but overall, the distance, and thus the need to queue, affects the residents’ exchange frequency to a certain extent, but not their attitudes. Residents who live farther away also have a positive attitude towards supermarkets. However, there is a correlation between residents’ perceptions of the cost of garbage exchange supermarkets and residents’ satisfaction. The analysis showed that the p-value between the two was 0.001, the p < 0.01, and the path critical ratio (C.R.) value was −3.286, indicating that cost perception was negatively correlated with satisfaction, assuming that H5 was confirmed. This may be due to the occasional need for queuing, the corresponding time required for garbage classification, and the distance from the exchange point affects the completion time of the transaction, which can affect the residents’ exchange experience and thus affect their satisfaction. Regarding the uneven distribution of supermarkets, which causes inconvenience for several residents, the organizers from Xixinan village pointed out that the garbage exchange supermarket is currently in a stage of rapid development. In the future, the initiative will be extended to every village as far as possible to meet the exchange needs of residents in different places and reduce the exchange cost.
Finally, the results of the path analysis show that the standardized coefficient of residents’ attitudes towards garbage exchange supermarkets and the path of environmentally responsible behavior is 2.240 and the p value is 0.0242, which is less than 0.05. Therefore, the positive effect of residents’ attitudes towards garbage exchange supermarket on residents’ environmentally responsible behaviors is confirmed. Thus, H3 is established. Findings also prove the relationship between residents’ satisfaction with garbage exchange supermarkets and their environmentally responsible behavior, with the path coefficient of 3.3598 and the p-value less than 0.001, indicating that satisfaction has a positive effect; therefore, H7 is confirmed. Furthermore, the relationship between attitude and behavior has been widely confirmed in different fields. In this study, D3 (I hope the supermarket can continue to run) and D4 (I support the establishment of a garbage exchange supermarket) can directly affect whether residents participate in the exchange. For the dimension of satisfaction, the management model of garbage exchange supermarkets and the types of exchange items can also affect the behavior of residents. Moreover, examining the behavior of residents in this study not only includes whether to go to the exchange, but also measures the derived environmental behavior, such as whether to persuade friends to participate in volunteering activities. Regarding the behavior of environmental responsibility, the residents in the field survey indicate that they are willing to collect garbage and exchange, which also shows their strong awareness of the need for environmental protection. Therefore, the establishment of the garbage exchange supermarket effectively influences residents to be environmentally responsible.