This article examines how psychometric models based on certain assumptions have come to be used counterproductively by many practitioners in ways that limit the kinds of conclusions that can be made. The general problem of the context's influence on performance is discussed, and some implications are drawn. (SLD)