*4.1. Descriptive Statistics*

It was uncovered that approximately 80% of the targeted individuals responded that they usually require less than three minutes (300 m) to arrive at the closest garbage collection spot. It was also revealed that more than half of the communities adhere to the rules on garbage sorting, and had installed garbage bins for separate disposal. Nonetheless, the proportion of respondents who consistently or almost frequently sort their garbage at home before discarding is less than 40%. Roughly 40% of the respondents revealed that they separately discard garbage at the collection spot. Among them, 55.6% confirmed that they are intellectually equipped with the correct knowledge of garbage sorting. In contrast, about 40% of the respondents were found to sort garbage based on incorrect knowledge (see Figure 3).

**Figure 3.** Behavior related to sorting household garbage. Source: authors' own calculations, 2020.

With respect to the garbage sorting program, 60% of the respondents asserted that they support it, 20% asserted that they somewhat support it, and only 1% opposed it. However, the proportion of respondents who agree or slightly agree to comply with the refuse charge system is 40% (see Figure 4). In other words, although 60% of the respondents are highly supportive of the garbage sorting program, an equal number (60%) of them are indi fferent to the refuse charge system.

**Figure 4.** Receptiveness to the refuse charge system. Source: authors' own calculations, 2020.

The survey also inquired into the environmental and moral value judgements of the respondents in relation to a 16-item scale of pro-environmental behavior and a nine-item scale of altruism (see Table 3). In creating the pro-environmental behavior scale, this study took its clues from a study by Zhao et al., and used a new approach that produces an accurate and reliable measurement based on the separate analysis of three aspects of pro-environmental behavior: purchase, use, and reuse [31].


**Table 3.** Pro-environmental behavior scale items.

Source: Authors' own calculations, 2020. Notes: SA is 'strongly agree', SWA is 'somewhat agree', U is 'unsure', SWD is 'somewhat disagree', SD is 'strongly disagree', ri-t is the item-total correlations, α is Cronbach's coe fficient alpha, and FL is factor loading. Percentages may not sum to 100 due to rounding.

The altruism scale was constructed as a new scale for this research. We applied the Schwartz norm-activation principles to measure altruistic attitudes (see Table 4). According to Schwartz's model, altruistic behavior arises from personal norms if two criteria are met: an individual must be aware that particular actions (or inactions) have consequences for the welfare of others (awareness of consequences, AC), and an individual must ascribe responsibility for the consequences of those actions to himself or herself (ascription of responsibility, AR) [32,33]. The simultaneous presence of AC and AR in a specific situation enables pertinent personal norms to motivate behavior.


**Table 4.** Altruism scale items.

Source: Authors' own calculations, 2020. Notes: SA is 'strongly agree', SWA is 'somewhat agree', U is 'unsure', SWD is 'somewhat disagree', SD is 'strongly disagree', ri-t is item-total correlations, α is Cronbach's coefficient alpha, and FL is factor loading. Percentages may not sum to 100 due to rounding.

Durkheim provided an eloquent analysis of the importance of moral norms in influencing human collective behavior [34]. Personal norms are perceived moral norms that represent personal beliefs about what is right and wrong when acting in a particular manner in a specific situation. Moral norms are a shared set of beliefs, values, and ideas on what is presumed to be the right behavior. They regulate social life or human affairs by guiding and restraining individual behavior and action that produces adverse consequences for other members of society. Thøgerson [35] discussed the role of social norms in restricting individualism in favor of collectivism; in 2009, Thøgerson [36] further explored the strength of a person's norms—namely, subjective social norms and personal norms—in guiding environmentally responsible behavior. Personal norms or moral norms inform our sense of identity and behavior. They represent the will of individuals to altruistically prioritize collective interest over self-interest.

The research presented here applies Schwartz's principles in the form of a general altruism scale based on Clark et al. [22]. The scale contains a total of nine items that test for the presence of individual personal norms, AC, and AR. Specific items are listed in Table 4. Items 1, 3, and 4 refer to personal norms; Items 2, 5, and 8 represent AC; and Items 6, 7, and 9 represent AR.

For the altruism scale, Cronbach's coefficient alpha was used to evaluate internal consistency. When all of the nine items were used, Cronbach's coefficient alpha was found to be 0.434. Given this, the items with higher values (namely, Items 3, 6, and 8) were eliminated from the analysis.

This study used a discrete choice model for the analysis. The objective variables considered are end-point garbage sorting behavior ('sorting behavior'), receptiveness to a refuse charge system ('receptiveness to fees'), and receptiveness to policies requiring garbage sorting ('receptiveness to policies'). The values of the garbage sorting behavior variable, an objective variable, were computed based on the responses derived from the questionnaires. These include, for example, "Do you sort garbage when you discard it at a collection spot in your community?" In answering the question,

the respondents are required to choose one of the five response options: always, very frequently, occasionally, rarely, and never. A five-point Likert scale was used to assign values to the responses (5 = always, ... , 1 = never). The variable 'receptiveness to fees' corresponds to the question, "Do you support the idea that people discarding more garbage should be charged a higher fee?" Likewise, the variable 'receptiveness to policies' is related to the question: "Do you support the garbage sorting system?" For these questions, the respondents were required to select one of the five response options: strongly agree, somewhat agree, unsure, somewhat disagree, or strongly disagree. Again, a five-point Likert scale was used to assign values to the responses as stated (5 = strongly agree, ... , 1 = strongly disagree).

As shown in Table 5, the explanatory variables include: (1) *EI* (Knowledgeable or not knowledgeable ('Agree or not agree' in Shenyang) of the local economic incentive program: yes = 1, no = 0), (2) *PEB* (the score of the pro-environmental behavior scale), (3) *ALT* (the score of the altruism scale), (4) *OWNERSHIP* (have or do not have a residential ownership: yes = 1, no = 0), (5) *MANAGEMENT* (the communities have or do not have a property management: yes = 1, no = 0), (6) *RULE* (the communities have or do not have the rules on garbage sorting: yes = 1, no = 0), (7) *INFRA* (the communities have or do not have the infrastructure to support garbage sorting: yes = 1, no = 0), (8) *DIST* (distance to the garbage collection spot: 5 = more than or equal to 10 min, 4 = 7–9 min, 3 = 4–6 min, 2 = 1–3 min, 1 = less than 1 min), (9) *KNOWLEDGE* (knowledge on the city's household garbage sorting regulations, derived from the answers to the question: 5 = very familiar, 4 = familiar, 3 = have heard about, but not familiar, 2 = never heard about the regulations, but have heard about the sorting instructions, 1 = never heard about the regulations or the sorting instructions), and (10) demographic variables including *GENDER* (male = 1, female = 2), *AGE* (above 60 = 5, 50–59 = 4, 40–49 = 3, 30–39 = 2, 18–29 = 1)*, MARRIAGE* (marital status, yes = 1, no = 0), *REGIS* (household registration status, native = 1, outsiders = 0), *EDUCATION* (education level: postgraduate or above = 4, undergraduate or junior college = 3, senior high or senior secondary = 2, junior high and below = 1), *OCCUPATION* (yes = 1, no = 0), *INCOME* (above 4001\$ = 8, 3001–4000\$ = 7, 2501–3000\$ = 6, 2001–2500\$ = 5, 1501–2000\$ = 4, 1001–1500\$ = 3, 501–1000\$ = 2, less than 500\$ = 1), *POLITICAL* (political affiliation, member of the CPC = 4, the Communist Youth League = 3, non-communist or minor parties = 2, commoner or others = 1), and *CITY* (Shanghai = 1, Shenyang = 2, Chengdu = 3).


**Table 5.** Description of the explanatory variables.


**Table 5.** *Cont*.

Source: authors' own calculations, 2020. Note: N = 1621, SD: Standard deviation, SD/%: this column refers to the standard deviation (SD) unless otherwise noted (%).

### *4.2. Estimation Results*

The results of the analysis are shown in Tables 6 and 7. Table 6 shows the estimation results for all of the respondents. The estimated coefficients on *PEB* and *EI* are statistically significant in the expected direction. More specifically, the positive signs on both variables indicate that the stronger the environmentally friendly activities and knowledge are, the higher the probabilities of participating in the garbage sorting program, the receptiveness to fees, and the receptiveness to policies are. The result supports the idea that economic measures exert a positive impact on inducing and promoting residents' garbage sorting behavior. They also serve to increase their receptiveness to the introduction of a refuse charge program, or to policies mandating garbage sorting. Nonetheless, the altruism scale (ALT) only has a significant positive correlation with receptiveness to policies. This indicates that altruistic individuals will engage in pro-environmental behaviors when there are environmental benefits. However, when it comes to private benefits, various responses were noted. For instance, in response to the question of "What charging method do you think is appropriate if the governmen<sup>t</sup> introduces quantity-based charging for garbage collection?", 416 respondents (the largest number) chose a designated bag system (Figure 5).



Source: authors' own calculations, 2020. Note: the symbols \*, \*\*, and \*\*\* indicate significance at the 10%, 5%, and 1% levels, respectively.

**Figure 5.** The refuse charge system. Source: authors' own calculations, 2020.

However, in response to the further question of "How much are you willing to pay if a designated bag system is introduced?", approximately 60% of the respondents indicated 0.01 to 0.02 yuan, which is, by any standard, surprisingly low (Figure 6). Implicitly, even though some respondents revealed that they support a refuse charge system, they are egoistically unwilling to bear any financial burden. This contradicts Schwartz's principles of altruism, in that the awareness of consequences (AC) did not lead to the ascription of responsibility. In other words, these respondents displayed a high self-interest, and are unwilling to engage in pro-environmental behavior or action in the presence of a perceived cost of sacrifice, as expressed in terms of a financial burden.

**Figure 6.** WTP (Willingness to Pay) for waste collection. Source: authors' own calculations, 2020.

The signs and coefficients for *AGE* indicate that the elderly are more likely to participate in the garbage sorting program, and display a higher receptiveness to fees. One reason for these tendencies could be that the young workers are unable to freely join sorting efforts that are scheduled during specific hours in the morning and evening in areas where the separate collection and disposal of HSW is implemented. There is also an asymmetric information issue. It may well be that campaigns to advertise the Green Account, the Green Earth program, the refuse charge system, and policies related to garbage sorting are mostly held during the daytime, with the consequence that the information fails to reach young workers. This may be one of the main factors that impede the wider recognition and adoption of sustainable HSW programs and policies or refuse charge systems among the younger population.

*GENDER* positively correlates with receptiveness to fees. That is, males were shown to be more likely to support a refuse charge system. The signs and significance of *OCCUPATION* indicate that people who have a stable job are more likely to become 'willing' participants. This may be attributed to the fact that businesses are more effective and quicker at educating employees on policies requiring garbage sorting and actual waste sorting. However, *OCCUPATION* does not have an effect on the receptiveness to fees and the receptiveness to policies. The signs and significance of *INCOME* show that the higher an income is, the higher the willingness to pay the refuse charge is.

Furthermore, the coefficients for *MARRIAGE* and *OWNERSHIP* are not significantly deviated from zero, indicating that neither affects the probability of participation, receptiveness to fees, or receptiveness to policies. The estimated coefficients on *EDUCATION* were found to be statistically significant in the negative direction of the garbage sorting behavior; that is, the higher a person's education level is, the less likely he or she will be willing to engage in garbage sorting. This result contradicts existing studies.

One the main reasons behind the above contradicting trend is that people with higher education levels tend to work longer. Hence, if sorted garbage is scheduled to be collected during specific hours in the morning and evening, as mentioned above, these people, in most cases, are not able to participate in HSW separation and collection. The signs and significance of *REGIS* indicate that the 'outsiders', who do not have a household registration for their area of residence, are more likely to be willing participants. They also displayed a higher receptiveness to policies. This result is in contrast to what was expected before the estimation. One of the main contributing factors behind this deviation is probably due to the 'sense of place' psychological factor of the 'outsiders', which leads to an increase of their feeling of identification, and to a positive change in their strength of relationship with the environment within which they exist. Choy [23,24] empirically established a strong relationship between the indigenous peoples' close attachment to the natural environment (sense of place) and their strong altruistic inclination for environmental protection, based on field research conducted in the tropical forest in Borneo, Malaysia. Choy [25] examined the values that the forest-dwelling indigenous people placed on the forested environment that they called home. He classified this as a 'sense of place' value. Semken [37] provides a good discussion of the strong sense of place of the American Indian and Alaskan Native people.

*POLITICAL* has a significantly positive correlation with behavior and receptiveness to policies, although its magnitude is small. It seems that members of the CPC display not only a stronger environmental awareness, but also a higher receptiveness to policies than the general public at large. One reason behind this pro-environmental inclination could be that, in the garbage separation pilot area, Primary Party organizations also play a key role in garbage sorting, and members of the CPC have acquired a green habit of actively taking part in garbage sorting [38,39]. This spontaneously sets an example to inspire and induce their neighbors and relatives to adopt environmentally friendly practices by introspection. This result also supports existing studies, such as Ghorbani et al. [21].

For matters governing the disposal of the sorted household garbage, the estimated coe fficients on *RULE* imply that communities which establish garbage sorting rules are more likely to participate in a garbage sorting program. They also revealed a higher receptiveness to waste managemen<sup>t</sup> policies. However, as regards *INFRA*, the existence of proper equipment at the garbage collection spot significantly and positively impacts garbage sorting behavior. The signs and coe fficients for *MANAGEMENT* indicate that communities with a property managemen<sup>t</sup> company show a higher level of receptiveness to policies. More specifically, in a community with a property managemen<sup>t</sup> company, the company undertakes the task of sorting the garbage on behalf of the residents. This reduces the burdens of the residents. Inexorably, this tends to induce a moral sense of obligation incumbent upon the residents to react positively to policies requiring garbage sorting. The coe fficients for *DIST* show that the more time required to arrive at the collection spot, the less receptive the individuals are to policies requiring garbage sorting.

Additionally, the estimated coe fficients on *CITY* statistically, significantly, and positively correlate with the receptiveness to fees and to policies. This implies that there are di fferences in receptiveness to fee and policy support among cities.

The estimation results for the respondents who perform end-point sorting are shown in Table 7. The results of the estimation are similar to those of the case with all of the respondents. However, the coe fficients for *MANAGEMENT*, *DIST*, *MARRIAGE*, and *POLITICAL* are not significantly correlated with the objective variables, indicating that none of them a ffect the probability of participation, receptiveness to fees, or receptiveness to policies.

The signs and significance of *INCOME* show that people with a higher income are more likely to take part in garbage sorting. *OCCUPATION* is also significantly and positively correlated with the receptiveness to policies. It seems that because businesses are quick to educate employees on policies requiring garbage sorting and actual sorting wastes, they have contributed to the enhancement of individuals' receptiveness to such policies. The estimated coe fficients on *AGE* are not statistically significant with respect to sorting behavior and the receptiveness to policies. On the other hand, *RULE* and *INFRA* have a significantly positive correlation with garbage sorting behavior. The estimated coe fficients on *KNOWLEDGE* were found to be statistically, significantly, and positively correlated with garbage sorting behavior, the receptiveness to fees, and the receptiveness to policies. It seems that the knowledge of the city's household garbage sorting regulations stimulates people to participate in the garbage sorting program. This has the e ffect of boosting their receptiveness to fees.

The estimation results show that garbage sorting behavior, the receptiveness to fees, and the receptiveness to policies significantly vary across the three cities. An analysis with propensity score matching was also performed. The receptiveness to fees and the receptiveness to policies were set as outcomes, the existence of economic incentives was set as an assignment variable, and the respondents' attributes were set as covariates. All of the coe fficients were negative. In other words, compared with the residents in Shanghai and Chengdu, where economic incentive measures such as the Green Account and the Green Earth program have been introduced, the residents of Shenyang, where such measures are non-existent, were found to engage more actively in garbage sorting. They are also more receptive to a refuse charge system and policies requiring garbage sorting.


Source: authors' own calculations, 2020. Note: the symbols \*, \*\*, and \*\*\* indicate significance at the 10%, 5%, and 1% levels, respectively.

### *4.3. Overall Estimate Outcomes and Policy Implications*

All of the study areas clearly show that many factors influence residents' sustainable waste disposal behavior and practice. These may be widely divided into the following categories:


Furthermore, the analysis based on the external variables indicates that the participants of the garbage sorting program tend to live in communities that have clear rules on garbage sorting and have installed the proper equipment at garbage collection spots. Additionally, the development of the solid waste managemen<sup>t</sup> infrastructure—such as smart garbage collecting stations that are designed to recognize different types of waste automatically—can encourage citizens to cooperate with the local governmen<sup>t</sup> in waste sorting.

The analysis also uncovered the interesting fact that participants of the garbage sorting program tend to be elderly and employed. In addition, the study revealed a lesser known, let alone well-analyzed, issue that the 'sense of place' can serve as a crucially important intrinsic impact factor in inducing an individual's green behavioral practices.

Our investigation also uncovered the important fact that the power of the external influence arising from politically influential and environmentally inclined elites, such as CPC members, is positively related to the state of green environmental consciousness or the green mentality of the individuals surrounding them. These findings reflect the combined significance of external and internal moderations, and the importance of advertising and educational activities with respect to the garbage sorting policies in each community. In the future, the local governmen<sup>t</sup> could optimize the use of metro or subway advertising media or social media such as WeChat or Tiktok in order to disseminate information concerning sustainable waste disposal practices. These critical measures also serve to induce communities to strengthen their waste sorting rules and set up their waste collection spots properly.

Our study further revealed that residents' receptiveness to a refuse charge system varies across cities, and many respondents tend to oppose the implementation of a refuse charge program. It was also found that, if a fee-based system were introduced, a designated bag system would be the most effective to draw support from the residents. In addition, compared with the residents in Shanghai and Chengdu, residents from Shenyang—where economic incentive measures have not been introduced are found to be more actively engage in garbage sorting. They are also more receptive to a refuse charge system and policies requiring garbage sorting. This implies that mandatory garbage sorting would be more e ffective than economic measures.
