**3. Materials and Methods**

*3.1. Data Collection*

This study took Hangzhou as the research area. Hangzhou is the capital city of Zhejiang Province, located in the south of China. The city covers a total area of 16,850 square kilometers and the local GDP is 1.61 trillion CNY (Chinese Yuan, namely 0.23 trillion US dollars) with a residential population of 11.936 million in 2020 (Hangzhou Municipal Bureau of Statistics, 2021, the data mentioned below in this paragraph area also from this). Hangzhou is devoted to developing a digital economy and achieving high-quality development, which needs a large amount of young talent. By the end of 2020, Hangzhou's talent pool expanded to 2.945 million people with an annual increase of 5.2%. The net inflow rate of talent and overseas talent ranked at the top in China, and, for 10 consecutive years, Hangzhou has been rated as being among the "Ten Most Attractive Chinese Cities for Foreigners" (Hangzhou Municipal Bureau of Statistics, 2021). Hangzhou is becoming one of the most dynamic cities in China and is therefore suitable for investigating the young talent settlement intentions and the influence of air pollution perceptions [46].

According to the World Health Organization, People aged 14 to 44 are classified as young people. So, in this study, "young talent" refer to those who have a Junior college degree or above and are under the age of 44. This study uses data gathered from a questionnaire survey conducted by the research team, from April to June 2018, at the Hangzhou Future Sci-Tech City, the Hangzhou Economic and Technological Development Zone, Hangzhou High-Tech Zone (Binjiang), and key office buildings in central urban districts, where the inflow of young talent to Hangzhou is more concentrated. The target population of the questionnaire survey was young talent working in Hangzhou. The survey adopted a stratified random sampling method. With support from relevant government departments, survey respondents were randomly selected from a list of enterprises and their employees. A total of 1200 questionnaires were distributed with 300 for each abovementioned region. Overall, 1089 responses were collected, and, of these, 102 were excluded due to incomplete or invalid data. The final valid number of questionnaire responses was 987. Table 1 provides basic information about the survey respondents. A total of 807 (81.8% of the total sample) had a bachelor's degree or above. This included 49 (5%) with a doctoral degree and 176 (17.8%) with a master's degree. Further, 702 (71.1%) had an annual income of CNY 80,000 or more and 761 (77.1%) were under 35 years old. Based on age, educational attainment, and income level, the survey sample mainly consisted of young people with high educational attainments.


**Table 1.** Sample of the respondents.

#### *3.2. Measurement*

The questionnaire's measurement scales were based on those used by previous related studies. A small-scale pre-survey and analysis were conducted before the large-scale survey. During this process, the research team communicated with the respondents comprehensively and used their input to improve the questionnaire in terms of reliability, validity,

readability, and semantic accuracy to avoid any possible ambiguities arising from terminology. Chinese was used as the common language and due attention was also paid to minimize information loss during translation in paper writing [47]. The questionnaire items used a five-point Likert scale, and each respondent answered based on their judgment. In the questionnaire, an answer of 0 corresponded to "no such problem", 1 to "average", 2 to "not serious", 3 to "not very serious", 4 to "quite serious", and 5 to "very serious".

Based on Li [48] and Wang and Han [24], this study measured air pollution perception using three items, "the severity of PM2.5 in Hangzhou", "the air in Hangzhou is gray", and "the air is smelly in Hangzhou". To evaluate place attachment, this study referenced the studies by Kyle et al. [49] and Williams and Vaske [50]. Seven items are used to measure place attachment, "I like the cultural heritage of Hangzhou", "the landscape of Hangzhou gives me a sense of belonging", "Hangzhou possesses all kinds of living facilities that I want", "I have a good time with my colleagues (neighbors) in Hangzhou", "the help provided by the people around me makes me feel very warm", "I often feel respected in my life", and "I am willing to make efforts to make Hangzhou become better". To evaluate urban settlement intentions, the scale, based on the settlement intention scale developed by Hu and Weng [51] was adjusted so that it included four questionnaire items: "I am willing to stay and live in Hangzhou for a long time", "I have not considered the idea of settling in other similar cities", "if I were to choose again, I would still choose to work and live in Hangzhou", and "if I have the opportunity, I would recommend my relatives and friends from other places to live in Hangzhou". Satisfaction of living in a city was evaluated using three items, "I am happy to be able to work and live in this area", "I am satisfied with the living environment in the city", and "I often feel spiritually happy living here". Simultaneously, drawing on the results from existing research, this study selected seven control variables: gender, age, education, time spent in Hangzhou, income, development expectations, and family and friends in Hangzhou. Among them, gender and education were dummy variables. Variable "1" represented male in the gender variable and having at least one family member or friend in Hangzhou in the family and friends variable. Regarding the education variable, an educational attainment of junior college was considered a reference point. Other variables, such as age, time spent in Hangzhou, income, and career development expectations, were considered as continuous variables.

#### **4. Results**

#### *4.1. Reliability and Validity Tests of the Questionnaire*

This study used Cronbach's alpha coefficient to test the reliability of the measurement's variables to ensure the reliability and validity of the questionnaire. The results showed that the reliability of the four scales of place attachment, residential satisfaction, air pollution perception, and urban settlement intentions were 0.868, 0.974, 0.912, and 0.934, respectively. All four results were greater than 0.7 and, thus, had good reliability [52]. Further, according to the method suggested by Fornell et al. [53], this study used AMOS 24.0 (software to analyze structural equation modeling) to conduct a confirmatory factor analysis (CFA) on the four main variables to calculate the square root of the average variance extracted (AVE) value of each variable. The discriminant validity of each variable was tested by comparing the square root of the AVE value of each variable with the correlation coefficient between the latent variables, as shown in Table 2. The square root of the AVE value of all the variables was greater than the correlation coefficients. This indicated a good discriminant validity among the variables.

In addition, this study conducted a structural validity test on the four variables. The results showed that all the factor loading values in the four-factor model (model fit indices: *X*2/df = 4.684, RMSEA = 0.078, IFI = 0.925, CFI = 0.918) (*X*<sup>2</sup> denotes chi-square test, which can assess overall fit and the discrepancy between the sample and fitted covariance matrices. df denotes model degrees of freedom. The chi-square value and model degrees of freedom can be used to calculate a *p*-value. Model is good fit if *p*-value > 0.05. RMSEA is an abbreviation for Root mean Square Error of Approximation. It is a parsimony-adjusted

index, which is good fit if RMSEA < 0.08. IFI is an abbreviation for incremental fit index with values greater than approximately 0.90. CFI is an abbreviation for comparative fit index, which is good fit if CFI ≥ 0.90) were significantly higher than the general recommendation of 0.4, indicating that the measurement items of each variable could be better aggregated and effectively reflect the same construct. The results also showed that the four-factor model substantially fit indicators better than the other factor models. In summary, the tests described above indicate that the data of this questionnaire have high reliability and validity.



Note: The numbers in bold form a diagonal, and the diagonal line demonstrates the square root of the average variance extracted (AVE) value, and below the diagonal is the correlation coefficient of each variable. \*\*\* *p* < 0.01 with two-tailed test.

#### *4.2. Testing the Main Effect*

Models 1 and 2 in Table 3 show that, after controlling the relevant variables, the independent variable of air pollution perception significantly impacts the dependent variable of urban settlement intentions. The *<sup>R</sup>*<sup>2</sup> changes significantly, supporting H1 (<sup>β</sup> <sup>=</sup> −0.077, *p* < 0.01). Particularly, it is worth pointing out that, among the control variables, family and friends in Hangzhou and career development expectations have a significant impact on urban settlement intentions. In other words, migrant talent with family and friends in Hangzhou are more willing to settle in Hangzhou permanently. Additionally, development expectations are also a key influencing variable of urban settlement intentions. Development expectations depend on one's judgment of future employment and development prospects; the better the expectation, the greater the cost of "giving up." In 2019, the added value of Hangzhou's core digital economy industry was CNY 379.5 billion, which increased by 15.1% compared with 2018. Contrastingly, the growth was 14.6% for the e-commerce industry, 13.6% for the Internet-of-Things industry, and 15.7% for the software and information service industry. Such growths resulted from the rapid development of high-tech industries, which provide a career platform for young talent and raises their development expectations. These improvements made Hangzhou one of the top cities in China, in terms of young talent inflow.

#### *4.3. Testing the Mediating Effects*

Models 3–6 tested the mediating effects of air pollution perception on residential satisfaction and further influence on the urban settlement intentions of young talent. Firstly, according to the results of Models 3 and 4 in Table 3, the control variables, such as age and career development expectations, had a significant influence on residential satisfaction and remained robust both in Models 3 and 4. After the inclusion of the key independent variables air pollution perception in Model 4, its effect on residential satisfaction was significant (<sup>β</sup> <sup>=</sup> −0.167, *<sup>p</sup>* < 0.001), with *<sup>R</sup>*<sup>2</sup> changing to 0.024, further enhancing the model's explanatory power. Therefore, Hypothesis H2a was supported. Further, intermediary variables were included in Model 5. The results showed that residential satisfaction had a significant influence on the urban settlement intentions of young talent (β = 0.530, *p* < 0.001). The model's explanatory power increased by 23.3%, based on the amount of change in *R*2; thus, supporting Hypothesis H2b. In addition, this study further examined the effect of perceived air quality and residential satisfaction on settlement intentions. Model 6 incorporated both the independent variable of air pollution perception and the mediating variable of residential satisfaction. The empirical results showed that residential satisfaction had a significant influence on settlement intentions (β = 0.532, *p* < 0.001), while the influence

of air pollution perception became insignificant (β = −0.012, *p* > 0.05). After adding both the mediating and independent variables, the independent variable in the model became insignificant. Contrastingly, the mediating variable remained significant, according to the evaluation method of Baron and Kenny [54]. This indicates that residential satisfaction plays a mediating role between air pollution perception and urban settlement intentions altogether, thus verifying Hypothesis H2.

To further test the mediating effects, this study conducted a bootstrap test using the PROCESS macro for SPSS/SAS developed by Hayes [55], and repeated the sample 5000 times. The results showed that the indirect effect of air pollution perception on urban settlement intentions through residential satisfaction was 0.0886, with a 95% confidence interval of [0.055, 0.124] and *p* < 0.001. According to the criteria proposed by Preacher and Hayes [56] for testing mediating effects, if the confidence interval of the indirect effect does not include 0, then the indirect effect reaches a significant level. The empirical results show that the exclusion of a value of 0 also confirmed Hypothesis H2.

**Table 3.** Testing main effect and mediating effects of residential satisfaction.


Note: Standard errors in parentheses; \* *p* < 0.10, \*\* *p* < 0.05, \*\*\* *p* < 0.01 with two-tailed test; all the regression coefficients were non-standardized. VIF, Variance Inflation Factor.
