Code For Neyman-Pearson Confidence Intervals with Arbitrary Levels Of Confidence for Paired Counting

Author

W. E. Potter

Abstract

Confidence intervals are helpful in describing results that lack precision. Furthermore, the use of Gaussian distributions to generate confidence intervals for low-level measurements can be in error because, while a sample can have negative net counts, it cannot have negative activity. A computer program written in C++ is given that automatically computes Neyman-Pearson confidence intervals of the form [0, ##.##] for arbitrary, specified levels of confidence. The code runs on a personal computer with inexpensive C++ software. The utilization and interpretation of values generated by the code is discussed. Also values generated for different levels of confidence are compared.

Meeting

This abstract was presented at the 34th Annual Midyear Meeting, "Radiation Safety and ALARA Considerations for the 21st Century", Automated Applications Session, 2/4/2001 - 2/7/2001, held in Anaheim, CA.

 
Index of Midyear Meeting Abstracts