4.2. Measurement model
We used the two-step approach as described by Anderson and Gerbing (1988). Firstly, we assessed reliability and conver- gent validity as shown in Table 1, and, secondly, discriminant validity, as illustrated in Table 2. To examine reliability, Cronbach’s alpha revealed that all constructs showed a value above 0.6 (the bar adopted by Bagozzi and Yi, 1988). To test for convergent validity, composite reliability (CR), factor loading, and average variance extracted (AVE) were examined. The measures are acceptable if an individual item loading is greater than 0.7, CR exceeds 0.7, and AVE is greater than 0.5 (Gefen et al., 2000).
In order to examine the discriminant validity of the constructs, this study used the Fornell and Lacker (1981) criterion, whereby the average variance shared between each construct and its measures should be greater than the variance shared between the construct and other constructs. As shown in Table 2, the correlations for each construct are less than the square root of AVE for the indicators measuring that construct indicating adequate discriminant validity.
This study use Harman’s single-factor test to exam whether common method bias is presented. Harman’s single-factor test was conducted to examine common method variance (CMV). If a substantial amount of CMV is present, either (1) a sin- gle factor will emerge from exploratory factor analysis or (2) one general factor will account for the majority of the covari- ance among the constructs. The unrotated exploratory factor analysis reveals the presence of three distinct factors with eigenvalue greater than 1.0, rather than a single factor. These three factors account for 66.306% of the total variance. The first factor explains 46.985% of the variance indicating that CMV in our study is not apparent.