Statistical vs Analytical Significance: Is My Model Good Enough?
Abstract
The credibility of M&S used in the system acquisition process is crucial to informed use of those models in support of system design, T&E and training. The key to affordable VV&A activities is to focus V&V efforts on the part of the model that's important to your problem. Analytical significance is the key criterion for M&S comparison with test data, where the comparison is based on acceptance criteria that are determined by analyzing your M&S application. By taking this approach we can go beyond tests of statistical significance to tests of whether the model is good enough for what the user needs it to do. V&V for its own sake is never cost-effective, but V&V conducted to support a specific problem focuses those activities on key analytical issues and makes the process efficient and cost-effective. So ultimately, the answer to the question, 'How much V&V do I need to determine if my model is good enough'? Is 'It depends on what you want the model to do'!
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2001
- Accession Number
- ADA405565
Entities
Organizations
- Naval Air Warfare Center Weapons Division