tween these two constructs. The variance-extracted estimates should be greater than the squared correlation estimate’’. Table 3 shows that the average variances extracted from any pair of factors were greater than the corresponding inter-construct squared correlation estimates. This test supported the discriminant validity of each construct.