Abstract:

Compared the psychometric properties of Internet-based ability and achievement tests with earlier data on analogous paper-and-pencil tests in two studies, expecting no differences. In Study 1, two deductive reasoning tests were adapted for Internet use and administered to 2,182 subjects. Correlations between the two tests differed for the two media, but after sample comparability was enhanced, differences disappeared. Two latent answer pattern classes were found, but they did not correlate with administration method. In Study 2, a preliminary business administration knowledge test was administered online to 615 subjects. Correlation of performance with biographical characteristics was low, but equal for both test formats. Additional scores from 85 business students revealed two latent answer classes that were not related to administration method. Low validity of the business achievement test is attributed to its low reliability. Overall, performance differences between samples are attributed to differences in education and knowledge, and further research is recommended.