Convergent Validity

As given in Table 28, the factor loadings of all items were greater than .6 and all were statistically significant at the .01 significance level, giving weak support for convergent validity. In the validity sample, factor loadings of all observable indicators were equal to or greater than .63. Evidence for strong support of the item’s convergent validity can be obtained by reviewing their squared factor loadings, which are desired to achieve the cut-off level of .5 or higher. Squared factor loadings for all but one item (Be6, X2 = .40) turned out to accomplish this threshold, as the majority of items demonstrated squared loadings in the .80s.

The validity of each first-order sub-dimension as an indicator of the higher-order construct was evaluated similarly. All five-factor loadings were considerably high, varying from .64 to .93 and significant (range of t-values: 12.43 to 20.94). Four out of five first-order constructs turned out to have a squared factor loading (y) equal to or greater than .53. In line with the findings of the reliability stage, the benevolence dimension turned out to represent a somewhat limited effect of eWOM trust from this perspective. Additional evidence for convergent validity is shown in Table 29, which includes the correlations between the components of eWOM trust which are all significant, varying from .47 to .83. Therefore, one is able to conclude that the five sub-dimensions all measure distinct aspects of the same construct.

< Prev   CONTENTS   Source   Next >