Definition
Model Credibility is a measure of the trust or confidence that a model’s results are acceptable for making decisions in a specific application. It is a subjective assessment supported by the objective findings of Verification and Validation.
Why It Matters
If stakeholders don’t trust a model, they won’t use it, regardless of its accuracy. Credibility is the ‘social license’ for technical work. Without it, even the most advanced simulations are ignored, leading to decision-making based on intuition and bias instead of data.
Core Concepts
- Relationship to V&V: Verification and validation are the primary technical processes that contribute to credibility.
- Accreditation: The formal recognition that a model has sufficient credibility for a specific purpose.
- Utility vs. Cost: Credibility is not absolute; its development requires an investment of resources. As credibility increases, the utility of the model increases, but often faces diminishing returns for incremental increases in cost.
- Contextual Nature: A model may be highly credible for one application (e.g., entertainment) but have zero credibility for another (e.g., flight qualification).