An Investigation of the Sensitivities of Neyer D-Optimal Testing to User Inputs and Test Population Characteristics

The sensitivity of the Neyer D-Optimal Sensitivity Test method to variation in user-defined inputs and properties of the test population has been investigated. Limitations imposed on the design of certain ordnance firing systems had driven more frequent Neyer testing during ordnance lot acceptance, and the rate of rejection had historically been much higher than the consistency and history of the production of those devices would indicate. A concern that the test operators may have been poorly selecting user-defined inputs required by the Neyer SenTest™ Software provided the impetus for an investigation into the sensitivities of those inputs, as well as parameters intrinsic to the test population that would by unknown to operators. The investigation has determined that the Neyer test method is effective at determining the firing stimulus mean for a population, regardless of user inputs or population characteristics. The investigation has also determined that the Neyer test method may produce significant errors in the population standard deviation, which has a profound effect on the calculations for maximum no-fire stimulus (MNFS) and minimum all-fire stimulus (MAFS) that ultimately drive lot acceptance. Significant variation in test results was found for parameters completely out of the control or knowledge of operators, like the order in which devices are selected for test or the population’s true standard deviation. An alternative test method has not been defined herein, but evaluation and discussion within industry is recommended.