Construction and Use of Tolerance Bounds Based on Binary Data to Assess Margin and Uncertainty

• Activity Number: 331
• Type: Invited
• Date/Time: Tuesday, August 6, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Alix Robertson*+ and Edward V. Thomas
• Companies: Sandia National Laboratories and Sandia National Laboratories
• Keywords: QMU ; binary response ; tolerance bounds

Abstract

Binary response testing is required in engineering applications that involve critical, go/no go performance such as the detonation of explosive devices. Maximum likelihood methods can be used to develop QMU estimates that place confidence limits on extreme quantiles of the distribution. However, although the QMU estimates are analogous in form to QMU limits from variables data, special considerations apply. This talk outlines the data analysis methods that are used, and presents an example to illustrate the information that is obtained. Simulation results from the literature are reviewed for guidance on sample size and methods to estimate extreme quantiles.