**Table 1.** Descriptive items.

#### **5. Results**

#### *5.1. Preliminary Study*

In order to ensure that the collected data were suitable for all factor analysis, the Kaiser-Meyer-Olkin (KMO) test was carried out. The KMO test result was 0.867, greater than 0.8, and the Bartlett's test showed a satisfactory result when *p* < 0.000, so the samples passed the reliability test, and factor analysis could be carried out.

#### *5.2. Exploratory Study*

In this paper, SPSS 21.0 was used for exploratory factor analysis (EFA) to extract principal component factors. As shown in Table 2, nine factors were obtained through principal component analysis, and the cumulative explanatory variance of these factors was 82.7%. In addition, Cronbach's α coefficient was greater than 0.8, and the sample data were good, which shows that these nine factors can be used to explain all measured items.


**Table 2.** Results of the exploratory factor analysis.


**Table 2.** *Cont.*

Note: please see Table 3 for the meaning of the abbreviations in Table 2.

#### *5.3. Confirmatory Study*

Confirmatory factor analysis (CFA) was used to verify the model. It can be seen from Table 3 that standardized load coefficients were all greater than 0.8, and were significant when the confidence was greater than 95%, and R2 was greater than 0.5. Through the confirmatory factor analysis test, it showed that the measurement item structure was good and that the model could be accepted completely (χ<sup>2</sup> = 216.144, df = 114, χ2/df = 1.896, RMSEA = 0.061, CFI = 0.939, GFI = 0.868, AGFI = 0.835, NFI = 0.880, TLI = 0.931).

**Table 3.** Parameter estimates of the confirmatory factor analysis.



**Table 3.** *Cont.*

The results show that the t-value of each item was higher than the critical value, the significance level was 0.05, and the R2 value of each variable was greater than 0.5. This is sufficient evidence of convergence efficiency. The discriminant validity was tested by comparing the mean variance (AVE) extracted and the square correlation between structures. The results show that the AVE was greater than 0.5, and the correlation coefficient between factors was less than AVE, indicating that there is good discrimination validity between factors (see Table 4).


**Table 4.** Discriminant validity.
