Uncovering Faking Samples in Applicant, Incumbent, and Experimental Data Sets: An Application of Mixed-Model Item Response Theory
使用混合模型项目反应理论，在求职者、在职者和实验数据中识别人格测验的作假子群，发现以往关于作假行为的假设过于局限。
Most research on faking personality inventories has assumed that individuals are either faking or responding honestly; distinctions within these two groups are generally not made. A recently developed statistical technique, mixed-model item response theory, was used to identify subgroups within samples of individuals taking two different personality inventories under various conditions. For one personality test, the authors obtained a sample of applicants and incumbents. For the second test, a sample of honest respondents and two samples of respondents instructed to fake (coached and ad lib) were obtained. Across the applicant and incumbent data sets, the authors generally found that three classes were needed to model all response patterns. In the experimental faking study, an honest class and an extreme faking class were needed to model the data. Overall, these results demonstrate that previous assumptions about the nature of faking on personality inventories have been too restrictive.