Model validation: philosophy, methodology and examples

Abstract

This paper proposes an integrated framework within which philosophical and methodological issues relating to model validation should be set. The viewpoint adopted is more expansive than that of traditional scientific rationality. Rather, it is suggested that there are several ways of 'dividing up reality', each of which has an associated rationality, dependent on purpose and problem context. Modelling methodology is then considered emphasising the role of validation 'sub-methodologies'. Examples are presented, chosen from the domains of physiology and medicine, which illustrate some of the major issues of model validation.