The authors have developed a performance test, CPTOP (Cognitive Performance Test of Productivity), which combines 11 basic task testss to evaluate intelligent abilities of office workers in order to quantitatively and objectively evaluate their productivity by controlling office environment. However, in the previous experiment, it was found that the results of CPTOP contain in-fluences of learning effect and variation of difficulties depending on each prob-lem set. In this study, therefore, subjective experiments have been conducted in order to compensate these influences and a correction method has been proposed.