2007 Volume 55 Issue 3 Pages 325-334
As an extension of the model of Toyoda, Kawahashi, and Matsushita (2005), the present authors added to the model the parameter f0, which represents individual differences in social desirability. In addition, we removed the assumption of tau-equivalent measurement for items. A paired-comparison test was constructed using competency scales developed in a pretest. The test was split into half, and the two halves standardized. The data from 400 participants indicated that split-half validity and reliability was confirmed. In a validity study, in order to examine the robustness of the standardized test to response biases, 199 participants took the paired-comparison test and a Likert test. Participants were instructed to assume that they were in a 2-selection evaluation scene. The results indicated that the new method is more robust to response biases than the Likert method, irrespective of differences in the evaluation scene.