**3. Methodology**

*3.1. Data and Samples*

This study employed a questionnaire survey to collect data. In July 2022, 500 questionnaires were distributed twice to Chinese managers and employees of enterprises; company

employees evaluated transformational leaders, and leaders reported basic company information. In total, 370 questionnaires were collected in October 2022, and the questionnaire recovery rate was approximately 74%. After eliminating invalid questionnaires, we finally obtained 350 valid questionnaires. The valid questionnaire recovery rate was 70%. From the recovered samples, 47% were male and 53% were female; 76.3% had a bachelor's degree or higher. Years in business was concentrated in less than 5 years category, which accounted for 34.86%. There were 87 state-owned enterprises, which accounted for 24.86%; 82 private enterprises, which accounted for 23.43%; 98 foreign enterprises, which accounted for 28.00%; and other businesses, which accounted for 29%. The finance and insurance industry employed 21 individuals, which accounted for 6%; the education industry employed 32 individuals, which accounted for 9.14%; and the transportation industry employed 26 individuals, which accounted for 7.43%. The manufacturing industry employed 41 individuals, which accounted for 11.71%, and services and 7 other industries accounted for 65.72%. Enterprises with 0–50 employees represented 23.14%, those with 50–200 employees represented 20.57%, those with 200–500 employees represented 32.86%, and those with more than 500 employees represented 23.43%. The demographic characteristics of the investigated respondents are shown in Table 1.


**Table 1.** Participants' demographic profiles.

#### *3.2. Variable Measurement*

The variables measured in this study were selected from established domestic and international scales, and the items were modified and adapted based on the actual situation. All scales were measured using a 5-point Likert scale, with 5 representing "strongly agree" and 1 representing "strongly disagree".

Transformational leadership: This study measures transformational leadership using a scale developed by scholars including Alrowwad and Chen [126,127]. The eight-item scale has demonstrated high levels of reliability and validity in previous research. Representative items on the scale include "the leader shows determination in accomplishing goals" and "the leader portrays an inspiring future". In this study, Cronbach's alpha coefficient for the scale is 0.895.

Organizational innovation: This study measures organizational innovation using a scale developed by scholars including Jansen and Zhou [128,129]. Previous research has demonstrated the reliability and validity of the five-item scale. Representative items on the scale include "the company introduces a new management system" and "the company introduces a new approach to planning and budgeting".

External social capital: This study measures external social capital using a scale developed by academics including Bornay-Barrachina et al. [130,131]. Previous research has demonstrated the reliability and validity of the six-item scale. Representative items on the scale include "The company maintains good relationships with government departments" and "The company maintains good cooperative relationships with its suppliers".

ESG performance: This study measures ESG performance using a scale developed by scholars such as De Roeck and Li [132,133], as well as data from databases such as Thomson Reuters. Environmental performance, corporate social responsibility, and corporate governance are the three dimensions of the scale. The environmental performance consists of six measures, such as "the company reduces environmentally harmful behaviors" and "the company uses clean energy and fuels". Corporate social responsibility consists of twelve measures, such as "the company values the welfare of its employees" and "the company participates in long-term social welfare activities". Finally, corporate governance consists of six measures, such as "the company has a good information disclosure mechanism" and "the company has good business ethics".

Control variables: Based on prior research, years in business, firm ownership, industry, and size were utilized as control variables. The number of employees determines the size of a company: 1 represents 0 to 50 employees, 2 represents 50 to 200 employees, 3 represents 200 to 500 employees, and 4 represents more than 500 employees. There are four types of firm ownership: state-owned enterprises, private enterprises, enterprises funded by foreign investors, and others. In addition, the firm industries are the financial industry, the real estate industry, and the service industry, among others. Construct and items are shown in Appendix A.

#### *3.3. Reliability and Validity Test*

Reliability test: The reliability test of the survey questionnaire is referred to as the reliability test. The alpha coefficient of Cronbach is primarily used for reliability tests. If Cronbach's alpha coefficient is greater than 0.7, each questionnaire item has a high degree of reliability. For example, in this study, Cronbach's alpha coefficients for transformational leadership, organizational innovation, external social capital, and corporate ESG performance are 0.895, 0.834, 0.882, and 0.880, respectively, all greater than 0.7, indicating that the questionnaire items for each variable have good reliability.

Validity test: Validity refers to the design and content accuracy of the questionnaire. As shown in Table 2, the variance explained and factor loading are used to test the validity of the questionnaire in this study. All KMO values exceed 0.8, indicating that the scale is appropriate for factor analysis. All variables in this study can be distinguished with precision. All the factor loadings of the questionnaire questions were greater than 0.7. This indicates that the questionnaire's convergent validity is high.


**Table 2.** Confirmatory factor analysis, reliability, and validity of measurement model.

#### *3.4. Confirmatory Factor Analysis*

We validated the primary variables of the study (transformational leadership, organizational innovation, external social capital, and ESG performance). According to the theoretical dimensions designed by the variable scale, ESG performance includes three first-order factors (corresponding to environmental performance, corporate social responsibility, and corporate governance, respectively). As shown in Table 3, the six-factor model assumed in this study has the best-fit index relative to other models, indicating that the six variables have good discriminant validity and correspond to six distinct constructs. The values of their χ2/df, TLI, CFI, RMR, and RMSEA were 1.380, 0.960, 0.962, 0.069, and 0.033, respectively, which were superior to those of the five-factor, four-factor, three-factor, two-factor, and one-factor models, indicating that the variables designed for this study had superior discriminant validity. As shown in Table 4, each construct's composite reliability (CR) was high, with the lowest value being 0.838, indicating that the constructs have good convergent validity. The study used AVE values for discriminant validity testing, and Table 4 shows that the AVE values of all variables are higher than 0.5, so it can be concluded that the variables have good discriminant validity.


**Table 3.** Results of confirmatory factor analysis.

Note: TL = transformational leadership, SC = external social capital, OI = organizational innovation, EP = environmental performance, CSR = corporate social responsibility, and GC = corporate governance. "+" indicates combined variables. Same applies below.

**Table 4.** Model AVE and CR indicator results.


#### *3.5. Multicollinearity Analysis*

To avoid serious correlations between variables, this study used the analysis of variance inflation factor (VIF) to determine if there is multicollinearity amongst the explanatory variables. An analysis of the results shown in Table 5 shows that the VIF values of each explanatory variable are below 5, with a tolerance greater than 0.1, indicating that there is no multicollinearity amongst the explanatory variables.

**Table 5.** Multicollinearity analysis.

