Campbelland Fiske (1959) suggest determining VPN

To further explore the three-factor model, data wereanalysed for male and female separately. Male andfemale sample sizes were considered adequate forexploratory analysis at this stage. All 17 items loadedas expected in both samplesÐsame factors emerged ineach case. In addition, the magnitude of the factorloading for all items was very similar. These resultsprovide evidence of reliability and construct validity forthe three dimensions of Internet self-e� cacy at theindividual level.Finally, the instrument's correlation matrix wasanalysed for convergent and discriminant validity.Using this approach, discriminant validity is tested foreach item by counting the number of times it correlatesmore highly with an item of another variable (factor)than with items of its own theoretical variable. Campbelland Fiske (1959) suggest determining whether this countis higher than half the potential comparisons. However,in this case, common method variances are present so itis unclear how large a count would be acceptable. Anexamination of the matrix for the 17-item instrumentreveals that only item 16 correlates more highly withitems of other constructs than with item 13 (an item ofits own construct). The correlation between items 16 and13 is 0.48 and signi®cant at p50.01. Each of theremaining 16 items is more highly correlated with theother items in its group than with any of the itemsmeasuring other variables. This suggests convergent anddiscriminant validity for the recommended three factor17-item instrument.