"The evaluation of selective devices merely by simple correlations with single criterion variables is insufficient. " Three dimensions of criteria that pose "embarassing and confusing questions" are discussed: Static dimensionality, or the problem of multiple performance measures that are not determined by the same general factors; dynamic dimensionality, or the problem of changes in criterion performance over time; and the criterion dimensionality of the individual[long dash]two individuals may be equally good, but for different reasons and in different ways.