Relationships between faking, validity, and decision criteria in personnel selection

There has been some debate in recent years as to whether faking on personality tests, while apparently not affecting criterion-related validity, still has a detrimental effect on the accuracy of hiring decisions. The present paper is set out to contribute to a clarification of this issue conceptually and empirically. In the conceptual part, statistical parameters of test scores obtained in selection settings that may affect validity and hiring decisions are disentangled. A data set of job incumbents who took an integrity test in a research setting is then used to demonstrate the effects of simulated faking scores with systematically manipulated distributional properties. Results show that, while hiring decisions are more sensitive to manipulations than validity, changes on both decisions and validity depend upon the same parameters, most importantly on variance in faking. Unlike the overlap between decisions based on faked and non-faked scores, the accuracy of these decisions was not more sensitive to faking than validity, regardless of selection ratio. Results are discussed in light of findings on criterion-related validity of personality tests in real-world applicant settings.