Chapter 10 - RETRACTED: How good (“valid”) are models?

Abstract

Abstract There remains considerable confusion in the scientific literature regarding the appropriate ways to evaluate how “good” a model is. The process of evaluating a model is without a doubt the most controversial step in the modeling process. This confusion arises more from semantics and differences in philosophical viewpoints than from debates over specific methodologies used. Regarding semantics, the term “validation” takes center stage, followed by terms such as “verification,” “confirmation,” “corroboration,” “credibility,” “qualification,” and “evaludation” (not a typo), which have been used in attempts to place “validation” within an appropriate context. Regarding philosophical viewpoints, some modelers have stated that “verification” and “validation” of numerical models of natural systems is impossible, but that their “confirmation” is possible. Oreskes et al. based their statements on an examination of the philosophical basis of these terms, thus their philosophical viewpoint is inextricably tied to their semantics. Other modelers have stated that “validation” simply means that a model is acceptable for its intended use because it meets specified performance requirements. Rykiel provided his statement while discussing the meaning of “validation” within the context of ecological simulation models. Still other modelers have coined the new term “evaludation” to avoid use of “validation” per se by combining it with the more neutral term “evaluation,” the intent being to keep the idea of “validation” visible while softening its connotation.