Established results on latent variable models are applied to the study of the validity of a psychological test. When the test predicts a criterion by measuring a unidimensional latent construct, not only must the total score predict the criterion, but the joint distribution of criterion scores and item responses must exhibit a certain pattern. The presence of this population pattern may be tested with sample data using the stratified Wilcoxon rank sum test. Often, criterion information is available only for selected examinees, for instance, those who are admitted or hired. Three cases are discussed: (i) selection at random, (ii) selection based on the current test, and (iii) selection based on other measures of the latent construct. Discriminant validity is also discussed.

We can make better decisions when we have a better understanding of the different sources of variance that impact job performance ratings. A failure to do so cannot only lead to inaccurate conclusions when interpreting job performance ratings, but often misguided efforts aimed at improving our ability to explain and predict them. In this paper, we outline six recommendations relating to the interpretation of predictive validity coefficients and efforts aimed at predicting job performance ratings. The first three focus on the need to evaluate the effectiveness of selection instruments and systems based only on the variance they can possibly account for. When doing so, we find that it is not only possible to account for the majority of the variance in job performance ratings that most select systems can possibly predict, but that we’ve been able to account for this variance for years. Our last three recommendations focus on the need to incorporate components related to additional sources of variance in our predictive models. We conclude with a discussion of their implications for both research and practice.