Operationally-relevant test lengths : a decision-analysis approach.

Abstract

This thesis approaches the question of How much testing is enough? by formulating a model for the combat situation in which the weapon (e.g., missile) will be used. Methods of Bayesian statistics are employed to allow the decision maker to benefit from prior information gained in the testing of similar systems by forecasting the operational gain from acceptance. A Microsoft Excel V7.0 spreadsheet serves as the user interface, and Visual Basic for Applications, Excel's built in macro-language, is the language used to produce the source code. The methodology accommodates two different tactical usages for the missile: a single shot, or a salvo of two shots. The missile might be acceptable if used in the two-shot salvo mode, but not in the single shot mode, and this would imply a greater cost per mission. In the end the missile might not be judged cost effective as compared to a competitive system. If the model proposed is (or can become) adequate much can be calculated/estimated before any operational tests are made. This could assist in economizing on operational testing