Calculating the Confidence That an Area Is Uncontaminated When Sample Results Are Subject to False Negatives

• Activity Number: 311
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Greg F. Piepel*+ and Brett Amidan
• Companies: Battelle/PNNL and Pacific Northwest National Laboratory
• Keywords: Anthrax ; Statistical sampling ; False negative ; Confidence ; Clearance ; Uncontaminated

Abstract

In 2001, letters containing Bacillus anthracis were processed through U. S. postal facilities and delivered to an U. S. Senate office building. In at least one case, samples were taken and the results were all negative, but the occurrence of anthrax disease and subsequent sampling led to positive results. A Government Accountability Office (GAO) investigation and reports identified the need for (1) validated sampling and analytical methods and (2) appropriate use of probabilistic sampling approaches. The GAO called for providing defensible statistical confidence statements about an area being uncontaminated (possibly after decontamination) when all sample results are negative. This presentation discusses the methods and formulas to address this problem when the results are subject to false negatives. The situations of presence-absence data and quantitative data are both covered.