Validation in Simulation: Various Positions in the Philosophy of Science

Abstract

There is still considerable doubt and even anxiety among simulation modelers as to what the methodologically correct guidelines or procedures for validating simulation models should be. Epistemically, the approaches one finds in the simulation literature run the gamut from objectivist to relativist with shades in between. At present in the philosophy of science, there appears to be a convergence toward a nonalgorithmic but discursive and nonrelativistic view of the argumentation involved in warranting scientific theorizing. The present paper attempts to give a description of the various philosophical positions as well as to summarize their problems and the kinds of evidentiary arguments they would each allow in arriving at defensible simulation models. From the debate, we attempt to set out a perspective that frees the practioner to pursue a varied set of approaches to validation with a diminished burden of methodological anxiety. Reciprocally this perspective does not let the modeler off of the hook but rather converts the validation problem into an ethical problem in which the practitioner must responsibly and professionally argue for the warrant of the model.