Detecting CPI Faking in a Police Sample:

Past research has suggested that person-fit indices computed using item response theory (IRT) offer promise with respect to providing a means for identifying individuals who produce aberrant responses on psychological tests. We examined the Z3 and F2 person-fit indices using the Responsibility scale from the California Psychological Inventory (CPI) in a sample of N = 393 State Trooper incumbents; each completed the CPI twice, once in an honest fashion, and once under instructions to reply as if applying for the job. As hypothesized, raters were successful at increasing their mean Responsibility scores in the applicant condition; however, significant differences between groups were seen only on the Z3 index, and ROC curves indicated that neither index effectively differentiated between conditions. These results suggest that IRT person-fit indices may be of limited usefulness when consistent dissimulation is likely (e.g., on transparent self-report personality tests). The fact that test validity was significantly moderated by rater-instructional set raises questions regarding the procedures that should be followed when conducting concurrent validation studies. The usefulness of item response theory (IRT) as a means for detecting faking or dissimulation of responses to personality instruments has been the focus of a number o