Modeling Effects of Faking on an Objective Personality Test

Abstract

The article investigated the effect of faking on the dimensionality and mean scores of an objective personality test measuring subjectively accepted level of risk. Various studies conducted with standardized personality questionnaires have demonstrated that faking may distort both the dimensionality and mean scores. In contrast, the empirical evidence on the effect of faking on the dimensionality and mean scores obtained with objective personality tests is still sparse. In the first study, we evaluated the effect of naturally occurring faking on the dimensionality and mean scores using a between-subject design that compared professional driver applicants to volunteers matched according to relevant sociodemographic characteristics. The results indicated that the objective personality test measures the same latent trait in both samples. Furthermore, we failed to observe a significant mean difference between volunteers and professional driver applicants. Furthermore, we failed to observe any mean score differences between both samples. The second study experimentally evaluated the effect of individual differences in respondents’ ability to infer the ideal level of the latent trait on the dimensionality and mean scores using a combined between- and within-subject design. Using modern item response theory techniques we found that a less clear description of the ideal level of the latent trait prevented a successful faking attempt. In contrast, a more clear and less difficult to implement a description of the ideal level of the latent trait allowed respondents to successfully increase their test scores. The results are discussed in light of current models of applicants response processes when working personality tests.