Does self‐face awareness influence green building project performance?—An empirical evidence from China

In the circumstances of “Carbon Peak,” it is imminent to implement green construction technology in the construction industry as it is a major source of CO2 emissions. Nevertheless, little study has been done on the impact of self‐face awareness (SA) on green building project performance (GP). On the basis of a large‐scale questionnaire survey in China and an in‐depth analysis of the four categories (cognition, behavior, satisfaction, and performance), this paper has added an SA dimension, and proposed a new impacting path. Then, based on the improved conceptual model for measuring the impact of green construction behavior on GP, this paper has conducted empirical tests by using a large sample of questionnaire data and partial least square‐structural equation modeling method. The results show that: (1) green construction behavior has a positive impact on GP, and the impact of perceived behavioral control of green construction technology is most significant for promoting green construction in the industry, and SA has a significant positive effect on green construction in the industry; (2) environmental perception and perceived behavioral control for green construction technology play a mediating role; and (3) significant differences indeed exist in the multigroup analysis, and the biggest difference is working experiences in the engineering field. Finally, this paper has proposed corresponding management implications based on the above findings. The main contribution of this study is that this study has developed a new dimension (SA) to the model considering Chinese national and cultural differences, in an attempt to reveal the impact of green construction technology on GP in the Chinese context. In addition, the promotion of green construction (PGC) in the construction industry was also added based on the sustainability perspective, which is the first attempt to explore the role of the PGC.


| INTRODUCTION
Green construction has been prevailing in the industry for achieving simultaneous development of economic, social, and environmental benefits of engineering projects. Specifically, green construction is conducive to save resources and reduce the negative impact on natural environment through scientific management and technological progress on the premise of ensuring quality and safety in engineering construction. Construction enterprises can implement green construction by choosing green materials, refining construction and minimizing resource input, strengthening resource recycling and reducing waste emissions, and so forth. At present, nevertheless, little research has been done on the impact of green construction behavior on green building project performance (GP) from a perspective of self-face awareness (SA).
Unlike traditional engineering projects, the goal of green building projects is the balance and integrity of economic, social, and environmental benefits. Green building 1 is one of the most effective solutions to improve engineering project performances through resource utilization and recycling, which can largely reduce the negative impact of the construction industry on the environment. Zhiyuan et al. 2 constructed a key performance indicator (KPI) evaluation index for green construction engineering project management based on four elements: econoSGT, technology, management, environment, and introduced the network analysis method (analytic network process [ANP]) to build a comprehensive evaluation model for green construction engineering project management performances using KPI-ANP and five projects in China's Qingdao construction enterprises for empirical analysis to verify the feasibility and rationality of the evaluation model. Pham and Pham 3 developed a framework of many hypotheses regarding interactions between green integration, environmental knowledge, and green performance using data from 19 projects in Vietnam. With the support of structural equation modeling (SEM), the findings show that environmental knowledge is a significant antecedent to the development of green integration strategies. Using questionnaires and SEM, Pham and Kim 4 verified that environmental, economic, and social practices have positive influences on the sustainability performance and leadership competences strengthening the environmental practices-sustainability performance relationship. Shen et al. 5 collected 87 feasibility studies and interviewed project managers and other construction industry staff to classify the attributes of project performance into three categories: economic performance, social performances, and environmental performance, and found that the economic performance was the issue that received the most attention, while the least attention was paid to social and environmental performances. Green buildings are more expensive than nongreen buildings, and it is necessary to determine their specific economic performance. Isa et al. 6 analyzed the factors influencing green office building investments in Malaysia using investors as the study population to determine the attractiveness of returns expected by investors in green office properties. The study shows that investors are starting to develop a strong interest in green office buildings because of their higher potential returns compared with nongreen buildings. Onubi et al. 7 studied the impact of environmental measures implemented during the construction process on the economic performance of 168 completed projects of Class A contractors in Nigeria using partial least square-structural equation modeling (PLS-SEM). The results indicate that contractors need to adopt flexible environmental measures to ensure a balance between economic and environmental performance. Rajendran et al. 8 studied the positive and negative impacts of green building design and construction process on the safety and health of construction workers. Collected data were analyzed to test whether there were differences in recordable incident rates and lost time work rates between green and nongreen projects. In addition, Olanipekun and coworkers [9][10][11][12][13] conducted analyses of key factors of project management environmental performances from a perspective of construction engineering firms and identified 19 controllable project management factors from relevant literature.
Existing literature review shows that the existing relevant research has made an in-depth study on the concept of green building technology (GBTS), and the application and understanding of GBTS have changed from a single dimension to a multidimensional measurement, and more elaborated researches have been conducted on exploring GBTS based on the concept of sustainable development. Nevertheless, empirical evidence is still very limited in the impact of green construction behavior on GP in the context of sustainable development. On the basis of the concept of sustainable development, this paper explores the interaction between environmental perception of green construction (EPG), basic knowledge of green construction (BKG), perceived behavioral control of green construction technology (PBC), promotion of green construction in the construction industry (PGC), and satisfaction of green construction technology (SGT) in a detailed way, quantifying the impact of green construction behavior on GP and how it can be tested with the model. On the other hand, given previous studies have not focused on the cultural context related to SA, this paper investigates the role of SA on the PGC and green building performance based on the specific cultural context in China.
This study intends to build a theoretical model for measuring the influencing mechanism of green construction behavior particularly SA on green building plant performance by adopting a structural equation model for empirical tests. To begin with, this study has designed and distributed a semistructured questionnaire based on the research hypotheses and structural equation conceptual model, and analyzed the questionnaire data using SPSS. Then, this study conducted a measurement model test, structural model test, mediating effect test, and multigroup analysis (MGA) by using PLS-SEM method and Smart PLS software. By conducting empirical tests on the integrated model, this study has made an attempt to reveal the influencing mechanism of green construction behavior particularly SA on GP to provide a theoretical basis for improving the efficiency of green building engineering projects. The main contributions of this study lie in the following: (1) This study has built a structural equation model for the impact of green construction behavior particularly SA on GP. This study investigated the relationship between BKG, PBC, EPG, SGT, PGC, and GP, by constructing a structural equation model based on the hypotheses proposed for this study. (2) This study has added a new dimension (SA) to the model in consideration of Chinese national and cultural differences, in an attempt to reveal the impact of green construction technology on GP in the Chinese context. (3) It has made the first attempt to explore the role of PGC. On the basis of an in-depth analysis of the four categories (cognition, behavior, satisfaction, and performance) related to the study, a new path towards GP is proposed and added, with an interesting finding: PGC has a direct positive impact on GP, which has been verified for the first time by the questionnaire data of this study.
The remainder of the paper is structured this way: Section 2 is the theoretical model and research hypotheses, Section 3 gives data collection and analysis, Section 4 presents the results, Section 5 analyzes the mediation effects and MGA, and Section 6 concludes the paper and proposes management implications. The framework of this paper is shown in Figure 1.

| RESEARCH HYPOTHESES AND CONCEPTUAL MODEL
In this study, we have proposed 11 hypotheses based on their relevant influencing paths.
Path 1: Basic Knowledge of Green Construction → Environmental Perception of Green Construction.
Human consciousness is the basis for guiding actions, and thorough understanding is often needed to carry out work smoothly. The same is true for the EPG. Currently, despite the fact that people's environmental awareness is getting higher, environmental awareness has not met the expected standards due to limited perceptions, resulting in little impact on environmental protection. Environmental perception is governed by ideology and consciousness, Zhao et al. 14 argued that people's EPG will change after they have basic knowledge about green construction, and most participants have better environmental perception ability and are willing to pay more for green construction than traditional construction upon F I G U R E 1 Framework of this paper. realizing the environmental benefits of green construction compared with traditional construction. Wu et al. 11 pointed out that intensive publicity of green buildings will enable the market more willing to pay for green buildings, which is beneficial to the development of green construction. Therefore, we propose the following hypothesis in this study: The existing studies point out that the motivation behind people's behavior is governed by perceptions, and that BKG will also affect how enterprises take green construction behavior. Ying 15 believes that the improvement of green awareness among employees in construction companies will lead to a better perception of green construction, which will influence the tendency of green construction behavior. Zsóka 16 believed that people's attitudes and willingness to act on green construction and environmental protection are more important in influencing behaviors. Liu 17 pointed out that there are also obstacles between low-carbon environmental awareness and behavior, these obstacles include regulatory and cultural aspects among others, these obstacles, and their mutual reinforcing effects hinder the process of converting cognitive awareness into behavior. Abidin 18 argued that the development progress of green construction behavior depends on people's perception and understanding of the consequences of their individual actions.
Therefore, we further propose the following hypothesis in this study: General environmental risk perception, pollution environmental risk perception, and technical environmental risk perception all contribute significantly to the public's environmental-friendly behavior. Wang et al. 19 pointed out that the environment awareness of GBTS developers has a significant positive effect on whether they adopt GBTS or not. Wu et al. 20 argued that the Chinese government is able to improve the environmental perception ability of contractors to show better C&D waste management behavior. Therefore, we propose the following hypothesis in this study: H3. Environmental Perception of Green Construction has a direct positive impact on Perceived Behavioral Control of Green Construction Technology.
Path 4: Environmental Perception of Green Construction → Promotion of Green Construction in the Construction Industry.
Environmental perception can also affect the promotion of green buildings. Dongning and Changhui 21 argued that the industry's reputation will be damaged by those companies with poor environmental perception, thus promoting industry associations to make relevant companies adopt behaviors to improve their environmental performance through various specialized ways. Wu et al. 20 believed that the immaturity of the market environment and the lack of environmental awareness are important factors for the slow development of cognition of green buildings in China. And the environmental awareness of different stakeholders has a significant impact on green building development. 11 Therefore, we further propose the following hypothesis in this study: H4. Environmental Perception of Green Construction has a direct positive impact on Promotion of Green Construction in the Construction Industry.
Path 5: Perceived Behavioral Control of Green Construction Technology → Promotion of Green Construction in the Construction Industry.
Peng et al. 22 found that improving green construction needs to strengthen the green awareness of staff and the standardized behavior of civilized construction so that green construction technology can be better applied in construction projects, thus promoting the rapid development of the construction industry. Kai 23 argued that there is a necessity and urgency to adopt green construction technology in construction projects, pointing out the significance and application strategy of green construction technology in China's construction industry. Therefore, we propose the following hypothesis in this study:

H5. Perceived Behavioral Control of Green
Construction Technology has a direct positive impact on Promotion of Green Construction in the Construction Industry.
Path 6: Perceived Behavioral Control of Green Construction Technology → Satisfaction of Green Construction Technology.
Zhao et al. 24 investigated the major obstacles affecting project managers' job satisfaction and proposed possible strategies to improve job satisfaction in green building projects. The study found that the perceptions of "job content" and "staff personality and competence" differ between green and traditional building projects. Meron and Meir 25 argued that the success of green construction projects is closely related to the performance of project managers, while job satisfaction plays an important role in the performance of project managers, and green construction perceived behavior control is closely related to the degree of satisfaction with green construction technology.
Therefore, we further propose the following hypothesis in this study:

H6. Perceived Behavioral Control of Green
Construction Technology has a direct positive impact on Satisfaction of Green Construction Technology.
Path 7: Perceived Behavioral Control of Green Construction Technology → Green Construction Performance.
The behavior of corporate green construction can also have a positive impact on corporate performance. Wei 26 believed that effective implementation of green supply chain management as a behavior can effectively improve the competitive strength of enterprises among their peers, enhance the organizational effectiveness of enterprises, achieve a reduction in construction costs for enterprises, and thus improve project performances. Kim et al. 27 considered that team performance depends on the coherent behavior of the entire team, not on the knowledge and skills of individual operators. Begum et al. 28 stated that the effective reduction of waste in the construction process can help improve the performance growth of the construction industry. Onubi et al. 7 found that the adopting green construction technology in construction projects improves the construction environment and is a means of economic operation. Environmental performance acts as a mediator between green construction behavior and economic performance of construction projects. Therefore, we further propose the following hypothesis in this study:

H7. Perceived Behavioral Control of Green
Construction Technology has a direct positive impact on Green Building Project Performance.
Path 8: Promotion of Green Construction in the Construction Industry → Satisfaction of Green Construction Technology.
Promotion of green buildings will promote technological satisfaction. Yu 29 found that the technical impetus comes from the technical environment in which the building construction enterprise is located. Applying water-saving, land-saving, material-saving, energy-saving, and environment-friendly technologies in the construction process can effectively reduce the consumption of engineering-related materials in the construction process, improve the environment and increase the owner's satisfaction with green construction technologies. Therefore, we further propose the following hypothesis in this study:

H8. Promotion of Green Construction in the
Construction Industry has a direct positive impact on Satisfaction of Green Construction Technology.
Path 9: Self-Face Awareness → Promotion of Green Construction in the Construction Industry.
Self-Face Awareness is conducive to promotion of green buildings. Ying 15 pointed out that face consciousness has a positive effect on the promotion of green buildings, and the research found that competitive pressure is one of the driving factors for corporate environmental management. Dongning and Changhui 21 showed that firms facing high public pressure were more likely to voluntarily adopt a standardized environmental management system, and the reason was face consciousness to reduce competitive pressure. Šelih 30 found that small-and medium-sized construction firms were reluctant to implement environmental management standards because they believed that no competitor was taking the lead, and the reason for this was also related to face consciousness. Therefore, we further propose the following hypothesis in this study: H9. Self-Face Awareness has a direct positive impact on Promotion of Green Construction in the Construction Industry.
Path 10: Promotion of Green Construction in the Construction Industry → Green Building Project Performance.
The development of any market does not happen overnight, and needs to be accumulated over a long period of time. In a carbon-reducing environment, the public's propensity to buy green products is also increasing, and market demand can largely accelerate the development of the green building market. Shen et al. 5 pointed out that project owners play a key role in influencing the sustainability performance of construction projects. Yu 29 argued that the more companies that adopt green construction technology, the more they can reduce construction costs, which will in turn make more companies to implement green construction technology. Therefore, we propose this new hypothesis:

H10. Promotion of Green Construction in the
Construction Industry has a direct positive impact on Green Building Project Performance.
Path 11: Satisfaction of Green Construction Technology → Green Building Project Performance.
There is a facilitative effect between technology satisfaction and performance. Chan and Chan 31 pointed out that the widespread use of energy-efficient building materials, regulation of the production environment, and reduction of waste generation can improve technology satisfaction and contribute to more targeted investments in environmental technologies, which can improve project performances, and that project success should also be measured by the satisfaction of project participants. Šelih 30 considered that customer needs to play a very important role in the development of green buildings. Zhang 32 pointed out that satisfaction has a positive and strong impact on project performance. Therefore, we further propose the following hypothesis in this study:

H11. Satisfaction of Green Construction Technology
has a direct positive effect on Green Building Project Performance.
On the basis of the above-mentioned hypotheses we proposed and the associated latent and measured variables, we have developed a structural equation model as shown in Figure 2.

| Questionnaire design
The questionnaire was designed based on a five-point Likert scale. The respondents were employees and managers of construction project teams including project investors and F I G U R E 2 Theoretical model of the impact of green construction behavior on green building project performance.
owners, design companies, construction corporations, consulting agencies, construction supervisors, and other construction-related organizations. The survey is divided into two major parts, the first part collected the basic information of the respondents and their companies. The second part addressed the measurement scale of seven dimensions included in the model: BKG, EPG, PBC, SGT, PGC, SA, and GP, with a total of 39 questions. The variable constructions and sources are given in Table A1.

| Questionnaire distribution and collection
The questionnaires are issued to the employees and middle and high-level managers of construction project teams including project investors and owners, design companies, construction corporations, consulting agencies, construction supervisors, and other construction-related organizations. In terms of data collection, considering the limitation of time as well as the convenience of answering the questions due to the influence of COVID-19 lockdown, the online "Questionnaire Star" platform was mainly used for our questionnaire distribution and data collection.
At the same time, to ensure the reliability and validity of the collected samples, the questionnaires were distributed in a nonpublic manner by inviting specific people to fill in the questionnaires. The sample with the same answer from the beginning to the end is considered invalid. A total of 446 questionnaires were distributed, 446 were returned, 39 invalid questionnaires were excluded, and 407 valid questionnaires were obtained, with an effective questionnaire rate of 91%. The distribution of sample information is shown in Figure 3.

| Descriptive statistical analysis
Descriptive statistical analysis gives the degree of concentration, discrete trends, and distribution characteristics of the sample data. From this, it can be judged that the data of this questionnaire are suitable for the next step of analysis. The detailed results of the descriptive statistics are given in Table A2.

| Common method variance (CMV)
Since all core variables in this study were reported by the subjects at the same time point, the sample data obtained may have CMV. To ensure the validity of the sample data, this study first tested this issue using Harman's oneway analysis of variance. All variables were put into an exploratory factor analysis and the results of the unrotated factor analysis were examined. If only one factor was analyzed or if the explanatory power of a factor was particularly high, a serious CMV was determined. As shown in Table A3, the test results show that the largest unrotated factor in this study can only explain 40.699% of the total variance (less than 50%), indicating that there is no serious CMV. 33,34 Next, this paper adopted the ULMC method to further test the CMV. Liang et al. 35 included in the PLS model a common method factor whose indicators included all the principal constructs' indicators and calculated each indicator's variances substantively explained by the principal construct and by the method. As shown in Table A4, the results demonstrate that the average substantively explained variance of the indicators is 0.709, while the average method-based variance is 0.012. The ratio of substantive variance to method variance is about 59:1. This analysis result further shows that the potential CMV in this sample data is within an acceptable range, and in-depth empirical analysis can be carried out.

| RESULTS
Structural equation models based on partial least squares (PLS-SEM) can estimate very complex models with many latent and explicit variables and can be applied to complex structural equation models containing a large number of structures, 36 with less stringent assumptions on the distribution of variables and error terms, and are more suitable for theoretical development. 37 Since the samples we collected and used for this study did not conform to a normal distribution, the covariancebased least-squares 38 method is not suitable for further analysis. PLS, however, does not require any normality assumptions, it is relatively better suited for nonnormal distributions. Therefore, the PLS-SEM is used in this study to test and evaluate the measurement model and structural model, for further analyzing the differences in different groups by adopting Smart PLS 3.3.3 software. The initial PLS-SEM model is shown in Figure 4.
According to Anderson and Gerbing, 39 we used a twostep approach to analyze and interpret the PLS-SEM results, which consists of two main parts: (1) external measurement model testing and (2) internal structural model testing.

| The measurement model
Wen and Li 40 proposed that the reliability and validity of the model should be assessed before evaluating measurement models. PLS-SEM is mainly used to confirm the reliability of the sample data from two aspects: firstly, it is the indicator reliability, which is evaluated by the Factor Loading; second, it is the internal consistency of the indicator data, which is mainly measured by the compositional reliability 41,42 and Cronbach's alpha (CA). The validity analysis mainly includes convergent validity and discriminant validity analysis, and the convergent validity of the model can be estimated by calculating the average variance extracted (AVE) 12 value.
The extreme difference values of the factor loadings ranged from 0.703 to 0.940, and the detailed results are illustrated in Table 1. The results of the reliability and convergent validity evaluations are given in Table 2, where CA values all ranged from 0.759 to 0.940, the AVE values ranging from 0.616 to 0.847, and the CR values were all greater than 0.800. CA value reflects the size of the scale's ability to explain the variables, and the internal consistency of the measurement results is considered high when the CA value is greater than 0.7. From Table 2, it can be seen that the CA value of each item is greater than 0.7, so the degree of internal consistency of this data set is high. The composite reliability (CR) is the degree of consistency of new potential variables in the model consisting of multiple variables, and the CR of each variable is considered high when CR is greater than 0.7. From Table 2, it can be seen that the CR value of each variable is greater than 0.7, so the degree of the CR is high.
AVE value responds to the authenticity and accuracy of the study. Fornell and Larcker 43 pointed out that 0.5 is the critical criterion for AVE, that is, to achieve a satisfactory level of convergent validity, the factor loadings of each measurement term need to exceed 0.5. The AVE values of the latent variables in this model are all above 0.5, indicating that the measurement model constructed in this study has good convergent validity.
To assess the discriminant validity, two methods were used in this study. First, according to the Fornell-Larcker criterion, 44 each latent variable has discriminant validity when the square root of AVE is greater than the correlation coefficient of each latent variable. Second, the cross-loadings of the measured items must be examined, and this method requires that each parameter has a greater factor loading in its own construct than in the other constructs.
The discriminant validity evaluations based on the Fornell-Larcker criterion are exhibited in Table 3. The diagonal bold text is the value of the root of AVE and the lower triangle is the Pearson correlation. It can be seen that the values on the diagonal are greater than the values on the horizontal or vertical columns, that is, the model's diagonal AVE root values are greater than the correlation of the relevant conformation.  A further basis for discriminant validity can be provided by examining the cross-loadings of the measurement items. Table 4 shows that each measurement item has the highest load in its corresponding structure and there is no cross-loading problem, indicating that there is no multicollinearity problem between items loaded in different components of the external measurement model. Therefore, the present model has good discriminant validity.
In summary, the measurement models constructed in this study all passed the test and were suitable for the next step of structural model testing.

| The structural model
Upon the validation of measurement models, we could further analyze the structural model. Structural models respond to causal relationships between potential variables. In this study, the tests of the structural model are divided into five parts: multicollinearity test, path coefficients test, coefficient of determination (R 2 ) test, predictive relevance (Q 2 ) test, and model fitness test.
The main indicator of the multicollinearity test is f 2 , which indicates the proportion of the entire model residuals explained by each additional conformation, reflecting the degree of influence of the exogenous latent variables on the endogenous latent variables. The f 2 values of 0.020, 0.150, and 0.350 indicate the predictor variable's low, medium, and large effect in the structural model. 45 As shown in Table 5, only PBC has an insignificant effect on PGC, while all other effects are medium and high. Therefore, as a whole, exogenous To test the paths assumed in the structural model, the path coefficients need to be evaluated. The calculation of path coefficients and significance was mainly performed using the bootstrapping of Smart PLS 3.3.3 software with a sample size of 5000 draws. The results are given in Figure 5. The values in Figure 5 are the path coefficients, and the corresponding p values are in parentheses. The critical value test criteria for the test are: p values within 0.001, 0.05, and 0.1 indicate highly significant, generally significant, and weakly significant, respectively.
Only the path coefficients of the two paths, PBC on PGC and SGT on GP, are generally significant at the 0.01 level. All other paths are highly significant, which indicates that all paths are supported in this model.
In addition, as can be seen from Table 6, the R 2 values of all latent variables range from 0.19 to 0.33, except for the R 2 of EPG, which is below 0.19. The R 2 values reflect the explanatory power of exogenous latent variables to endogenous latent variables, and R 2 values of 0.190, 0.330, and 0.670 indicate that the explanatory power of the model is low, medium, and high. 40 Overall the explanatory power of the model is between medium and large.
The main function of the Q 2 test is to evaluate the predictive relevance of the model, which is proportional to the Q 2 values. When Q 2 takes a value greater than 0, the model has predictive relevance, and the larger the value the stronger the predictive relevance. In this study, the predictive relevance of the model was assessed by cross-validity redundancy. Since in this study we used Smart PLS 3.3.3 software to test Q 2 values with the Blinfolding method, and the final cross-validity redundancy was obtained through the calculation of this method as shown in Table 7. According to the data in Table 7, it was found that the cross redundancy of all latent variables was greater than 0, which indicates that the predictive relevance of the adopted model is good.
Due to the distribution-free nature of the PLS method, it is not suitable to use the parameters of Covariance Base SEM to measure the fitness of the model. As an alternative, Tenenhaus et al. 46 proposed the geometric mean of the mean-variance extracted and the mean coefficient of determination to test the overall goodness of fit and explanatory strength of the model. When goodness-of-fit (GOF) indicator value is between 0.1 and 0.25, the explanatory power of the relevant model is poor; when GOF indicator value is between 0.25 and 0.36, the explanatory power of the relevant model is average; when GOF indicator value is greater than 0.36, the explanatory power of the relevant model is good. VAF > 80 represents fully mediated. Abbreviations: BKG, basic knowledge of green construction; EPG, environmental perception of green construction; GP, green building project performance; PBC, perceived behavioral control of green construction technology; PGC, promotion of green construction in the construction industry; SA, self-face awareness; SGT, satisfaction of green construction technology; VAF, variance-accounted-for.

Source:
Calculated by the authors based on questionnaire data.

Source:
Calculated by the authors based on questionnaire data.
Therefore, it can be seen that the structural model has a good GOF.
In summary, the structural equation model constructed in this study passed the measurement model test and the structural model test.

| Mediation effects
Variance-accounted-for (VAF) is an assessment of the size of the mediating effect that explains the proportion of variance. In this study, the mediation effect in the model is tested by calculating VAF.
The effect of BKG on PBC is partially realized through EPG, and EPG accounts for 34.70% of the total effect.
The influence of EPG on PGC is partially achieved through PBC, and PBC accounts for 23.90% of the total effect.
The detailed results of the test are shown in Table 8.

| Multigroup analysis
In this study, MGA is used to test the moderating effect of categorical variables. MGA refers to dividing the variable data into several subsamples according to certain categorical classification criteria, and then using the multimodel comparison method to test the significance of the differences between groups. The novelty of PLS-based multicluster analysis method is that it uses bootstrap random sampling for multiple iterations, and the test results are more stable. The confidence interval method for significance testing is usually chosen to correct for bias and accelerated Bootstrap (5000 subsamples) for a twotailed test (significance level set to 0.05).
The purpose of this test is to determine the differences in the test results of the model in different groupings. The categorical variables (nine categories in total) tested were gender, marriage, age group, type of company, education, years of work experience in the engineering field, job position, years of work for green buildings, and number of green building projects engaged.
For H7: Perceived Behavioral Control of Green Construction Technology → Green Construction Performance. In this path, there are significant differences between career and private, career and state-owned, and between working in the engineering field for less than 10 years and working in the engineering field for more than 10 years. For H10: Promotion of Green Construction in the Construction Industry → Green Construction Performance. In this path, there is a significant difference between engaging in 3 or fewer green building projects and more than 3. For H11: Satisfaction of Green Construction Technology → Green Building Project Performance.
In this path, there is a significant difference between the institutions and the private sector.
The final model of the impact of green construction behavior on green building project performance. XIE ET AL.

| 1975
No significant differences were found for any other paths except for the above-mentioned four. The detailed results of the test are given in Table 9.
The final influencing model of GP in this paper is derived, as shown in Figure 6.
6 | CONCLUSIONS AND IMPLICATIONS

| Conclusions
Through the above empirical analyses and tests, we have drawn the following conclusions.
(1) Green construction behavior has a positive impact on GP, and the impact of PBC is most significant for promoting green construction in the industry, and SA has a significant positive effect on green construction in the industry. Among several influencing factors that affect GP, PBC has the greatest impact on green construction performance. In the path of PBC on GP, the path coefficient was 0.662 and significant at the level of 0.001 and f 2 was 0.760, indicating that PBC has a greater impact on GP. Therefore, we argue that the stronger PBC, the higher the performance of green building projects. (2) EPG and perceived behavioral control for green construction technology played a mediating role in the influence mechanism on GPs. As a mediating variable between BKG and PBC, the improvement of EPG contributes to the influence of BKG and the enhancement of PBC, which in turn promotes the improvement of GP. As a mediating variable between EPG and PGC, improvement of PBC helps the BKG exert impacts and increase the promotion of green construction in the industry, which in turn contributes to the improvement of GP.

| Management implications
Green construction is an inevitable trend in the transformation of the construction industry in the context of carbon emissions reduction. The concept of green construction has been promoted in China for the past decades, but the development status is not optimistic so far. The results of this study can enable policy-makers to develop appropriate green construction promotion strategies. On the basis of the findings of this study, we propose the following management implications.
(1) The government should actively promote the concepts related to green construction, encourage the development of green construction, and vigorously promote the popularity of green construction. The research results show that PBC has the greatest degree of influence on green construction performance. By improving the BKG, PBC can be enhanced thereby improving GP. By enhancing training and education and creating credible green construction demonstration projects, public awareness, and perception can be increased, which has a positive effect on GP. (2) It is suggested that the government can further promote green construction by strengthening the regulations and implementation of green construction, actively advocating the theory of circular economy, and combining these with necessary incentives. Our study indicates that EPG can play an obvious mediating role. Therefore, by increasing the sensitivity of the social environment to green construction, GP can be improved and development of green construction can be promoted accordingly. (3) Enterprises need to provide training to their staff on green construction-related knowledge. This study finds that there are differences between groups with different years of work in different engineering fields, types of enterprises, job positions, and the number of green construction projects they are engaged in. Therefore, GP can be significantly improved by freshening up green construction knowledge with the staff of the construction project enterprises via necessary internal training programs.
government departments for their support and assistance during data collection in the field survey. The authors declare no conflict of interest. BKG7 I see no difficulty in the implementation of the environmental requirements of the Green Construction Code in the project.

Satisfaction of green construction technology (SGT)
SGT8 I think the use of green construction techniques will increase the satisfaction of the owner side.
Xinxu, 50 Heng et al., 51 and Xiaoling et al. 52 SGT9 I think the owner is satisfied with the results of the implementation of the green construction project.
SGT10 I think the owner side is more satisfied with the green construction project than the general construction project. SGT11 I think owners will be more inclined to choose green construction projects in the future.
Perceived behavioral control of green construction technology (PBC) PBC12 Adoption of Green Construction Technology-I think the equipment and construction materials required by green construction technology can be satisfied.
Gelhard and Von Delft, 53 Zhao et al., 54 Yuan et al., 55  GP24 I would like to make the construction process more environmental-friendly and green.
GP25 I believe that the adoption of green construction techniques contributes to the success of construction projects.
GP26 I believe that the adoption of green construction techniques contributes to the smooth performance of construction contracts.
GP27 I think the green construction will help the project receive the social award honor.
GP28 I believe that the adoption of green construction techniques will improve the performance of construction projects.
GP29 I think there are worthwhile practices for green construction projects.
Self-face awareness (SA) SA30 Green Construction Self-Face Consciousness-I care a lot about the compliments and praise from others.
Yu et al., 59 Xu, 60    Q24 I would like to make the construction process more environmentally friendly and green.

Source:
Calculated by the authors based on questionnaire data.