Utilizing Statistical Inference to Guide Expectations and Test Structuring During Operational Testing and Evaluation

Abstract

Comparative tests are commonly used during the operational testing phase to baseline the system under test (SUT) against the current status quo. Depending on the type of SUT, the comparative test may be costly and resource intensive. Thus, any insights which may be gleaned about the potential results of the test beforehand may provide guidance on (1) the potential benefits of conducting the test and (2) the structuring of the test. This paper offers a statistical approach to understanding the type of results which may emerge during comparative testing of the SUT. Specifically, we utilize the concept of statistical inference to determine the needed performance difference between the SUT and the baseline system. If performance differences are statistically different, there may be useful information to be gained from conducting the test as is. Performance differences, which are not statistically different, may indicate that the test should be restructured or postponed. In either case, the relevant decision-maker is provided with information about the potential results of the test beforehand in order to make an informed decision. We illustrate the method of statistical inference on a system which improves situational awareness on the battlefield.