Internet usage may be overrepresented

Self-e cacy is a domain dependent construct, therefore selecting items that are representative of a particulardomain is vital to ensure the content validity of the scale.The authors are aware that some aspects of Internet/Webusage are not represented in this instrument (e.g. browsercon®guration, downloading software, posting to newsgroups). It is also true that a few items, while related toInternet usage, may not be speci®c only to the Internetdomain (e.g. scanning a picture, sending a fax). Moreover, some elements of Internet usage may be overrepresented in the scale (e.g. encrypt/decrypt e-mail).This suggests that the content validity of the recommended instrument can be improved.While this suggests that the content validity of theinstrument can be improved, it is important to note thatthe recommended instrument is useful in its presentform. The results of the factor analysis indicate that the®rst factor (browsing) is the key to Internet usage andaccounts for 59% of the variance in self-e cacy scores.While the other factors account for substantially lessvariance, 13.6% for encryption/decryption and only7.9% for system manipulation, they are both highlycorrelated with the key (browsing) factor.Selecting self-e cacy items for new and ever-changingtechnologies like the World Wide Web will always beproblematic because items may become quickly dated.However, we cannot let this fact deter us from makingthe attempt. The current instrument demonstrates goodconvergent and discriminant validity along with highreliability. Although the content validity of the scale canbe improved, the current version will serve both as auseful instrument and a starting point for an improvedmeasure of Internet self-e cacy.