Koreascholar

4.2. Measurement Model 
The Confirmatory Factor Analysis (CFA) was employed to 
establish confidence in the measurement model with maximum 
likelihood (ML) estimated because it is consistent and 
asymptomatically efficient in large-scale samples as shown in 
Table 1 (Bollen, 1989). Proper evaluation of the measurement 
model is a prerequisite of the structural model, with the 
convergent validity of the measurement scale examined in 
terms of factor loadings and average variance extracted (AVE). 
According to Hair et al. (2014), convergent validity requires a 
factor loading that is greater than 0.60 and AVE not less than 
0.50. Table 2 shows that the majority of the indicators had 
significant factor loadings higher than 0.60 (p<0.05). However, 
five indicators, namely, TSQ2, TSQ3, TSQ6, TPA7, and TS8, 
were removed because the factor loadings were below 0.60. 
The average variances extracted ranged from 0.559 to 0.668, 
which showed strong convergent validity, therefore, the AVE 
surpassed the threshold value of 0.50. The Cronbach’s alpha 
values for all constructs showed strong internal consistency 
ranging from 0.870 to 0.907, while the construct reliability 
(CR) values were above the suggested standard of 0.70. 
Therefore, it can be concluded that all latent constructs possess 
sufficient reliability. According to Fornell and Larcker (1981), 
discriminant validity can be established when the AVE values for 
the latent constructs are compared with the squared correlations 
between the corresponding constructs, with none of the squared 
correlations surpassing the AVE. These tests indicated that the 
discriminant validity was upheld for all constructs.
4.3. Structural Model
The structural model involves the significance tests used 
to estimate coefficients (paths), which provide the basis for 
accepting or rejecting the proposed relationships between 
latent constructs (Chi & Qu, 2008). Prior to the estimation 
of path coefficients, a structural model with five constructs 
was estimated using ML estimation, as shown in Figure 2. 

---

Filda RAHMIATI, Norfaridatul Akmaliah OTHMAN, Mohammed Hariri BAKRI, Yunita ISMAIL, Grace AMIN /
Journal of Asian Finance, Economics and Business Vol 7 No 12 (2020) 959–968
964