Determination of confidence limits for experiments with low numbers of counts

Ralph Kraft, D. N. Burrows, J. A. Nousek

The Astrophysical Journal · 1991 · 488 citations · 0 references

DOIFull text

Open access

Concepts

TL;DR

Confidence limits for Poisson‑distributed data are crucial in low‑count experiments, and this study compares classical and Bayesian approaches. The study focuses on scenarios where observed counts are small and comparable to background counts. The authors argue that the Bayesian method is preferable and provide tables of its confidence limits for quick reference. Published in The Astrophysical Journal, June 1991, DOI 10.1086/170124.

Abstract

view Abstract Citations (405) References (5) Co-Reads Similar Papers Volume Content Graphics Metrics Export Citation NASA/ADS Determination of Confidence Limits for Experiments with Low Numbers of Counts Kraft, Ralph P. ; Burrows, David N. ; Nousek, John A. Abstract Two different methods, classical and Bayesian, for determining confidence intervals involving Poisson-distributed data are compared. Particular consideration is given to cases where the number of counts observed is small and is comparable to the mean number of background counts. Reasons for preferring the Bayesian over the classical method are given. Tables of confidence limits calculated by the Bayesian method are provided for quick reference. Publication: The Astrophysical Journal Pub Date: June 1991 DOI: 10.1086/170124 Bibcode: 1991ApJ...374..344K Keywords: Bayes Theorem; Confidence Limits; Photon Density; Poisson Density Functions; Supernova 1987a; Charge Coupled Devices; Error Analysis; Gamma Ray Astronomy; Probability Distribution Functions; STATISTICS AND PROBABILITY; GAMMA RAYS: GENERAL; NUMERICAL METHODS; X-RAYS: GENERAL full text sources ADS | data products SIMBAD (1)