ABSTRACT
In contrast to chapters 8 and 9, which focus primarily on predictor constructs, this chapter considers many kinds of predictor measures (including interviews, situational judgment, assessment centers, and work samples). These predictor methods are frequently called performancebased tests or simulations. We also consider the nature of physical ability tests, which are yet another variety of performance-based tests. As we saw in chapter 7, these various performance-based tests and simulations tend to have moderate to high criterion-related validity and favorable applicant reactions. This should come as no surprise because they often require applicants to perform the same (or highly similar) set of behaviors as those required on the job. For this reason, they have higher physical fidelity than traditional cognitive ability and personality tests. They also tend to produce smaller racial and gender subgroup differences than these other predictors (the exception is physical ability testing). However, they have two general limitations. The first is that many simulations are costly to develop and administer. The second is that, historically, they do not produce homogeneous scores for underlying psychological constructs. However, we see this is starting to change
because research conducted over the last several years has begun to focus on enhancing the construct validity of these performance-based predictor measures.
