The ratio ofsample size to number of items VPN

Seven of the 24 items shown in the appendix wereeliminated because they had a corrected-item totalcorrelation less than 0.50. There are no establishedcriteria for this cut-oŒ point. Researchers have frequently used a range between 0.5 and 0.7. Reliabilityanalysis supported these items. No item was deletedbecause it would improve reliability. Next, an exploratory factor analysis was conducted using principalcomponents as the means of extraction and varimax asthe method of rotation.Without specifying the number of factors, the factoranalysis of the seventeen items suggests a three-factormodel with eigenvalues greater than 1.0. The ratio ofsample size to number of items (16.29:1) was greaterthan accepted standard (10:1). The factor structure waseasy to interpret corresponding with browsing (threeitems), encryption/decription (six items) and systemmanipulation (eight items). The three factors explained80.6% of the variance. The range for factor loading was 0.78 to 0.87 for browsing, 0.81 to 0.91 for encryption/decryption, and 0.69 to 0.85 for system manipulation.Table 1 reports the simple three-factor solution (loadings greater than or equal to 0.4 are reported). Table 1also provides factor correlation matrix, means andstandard deviations for the three factors.Using this sample of 277 responses, the corrected-itemtotal correlation and reliability (alpha) for each of thethree self-e� cacy factors were calculated. The range forcorrected-item total correlation was 0.81 to 0.89 forbrowsing, 0.85 to 0.96 for encryption/decryption, and0.72 to 0.84 for system manipulation. Coe� cient alphareliability scores were 0.93, 0.98 and 0.94 for browsing,encryption/decryption and system manipulation, respectively. Overall reliability for the 17-item scale was 0.96.