Techniques for Reducing Applicant Response Distortion: Their Effects on Measurement Equivalence and Criterion-Related Validity
A challenge to the use of non-cognitive tests in selection is the possibility that applicants might distort their responses. Several techniques have been suggested to reduce response distortion, including the provision of warnings that faking is detectable, and the requirement that respondents elaborate on their answers to increase the difficulty of lying. In this study, the provision of warnings and requests for elaborations on responses were assessed for their potential impact on the factor structure, means, and criterion-related validity of a performance predictor. Results indicate that warnings improve criterion related validity relative to a control condition, but that elaborations distort factor structure. More specifically, participants appeared to interpret personality assessment items equivalently between the warning and control conditions, their scores were lower for those in the warning condition, and criterion related validity was significantly improved by the use of warnings. These results are promising for practitioners who use warnings in order to reduce faking on non-cognitive tests. In contrast, this study’s results suggest that elaboration manipulations should be used with caution. Elaboration requirements resulted in disrupted factor form/factor structure, which prevented the interpretation of mean differences between the experimental and control conditions.
Journal of Knowledge and Human Resource Management, Vol. 1, No. 1, 2009, 47-61.