Abstract
Changes in views of personnel selection research include changing opinions about choosing predictors and criteria, methods of data collection, and methods of evaluating selection procedures. In particular, distinctions are (or can be) made between job‐relatedness, inferred in part from validity coefficients based on job‐related criteria, and the valid measurement of a trait (attribute), inferred in part from research using construct‐related criteria. The major change identified is a reduction in the sanctity with which a single correlation is viewed for either purpose.