Comparing the variance extracted (AVE) estimates for each construct with the squared inter-construct correlations between the relevant constructs gives an indication of discriminant validity (see Table 3). The results in Table 3 show that the AVE is greater than the squared correlation estimate for each construct pair, providing additional support of discriminant validity .
Common method variance (CMV) is also assessed. A worse fit for a one-factor model with all items loading on a single construct suggests that CMV does not pose a serious threat. In this case, the one-factor test yields a χ2 of 170.6 with 90 df (pb<0.001). Obviously, the one-factor fit is much worse than the proposed measurement model (difference in χ2=60.5, df=10, pb<0.0001).