ABSTRACT
This paper was prepared as the stage-setter for a session on measuring performance of Department of Defense models of human, social, and cultural behavior. It is, then, about validation and usefulness-complicated issues when dealing with social-science models. I discuss these complications and offer provocative suggestions, which include recognizing: the range of purposes for which the models are used, that validation should reflect the style of inquiry, that both theoretical and empirical information should be used for validation, that empirical testing should be theory-informed and theory-building oriented, and that more diverse types of empirical information should be used. Finally, I see value in validating the qualitative structure of uncertainty-sensitive systemic models. These can then be used for exploratory analysis in search of conclusions robust across uncertainties.
