Methods to find a confidence interval for Poisson distributed variables are illuminated, especially for the case of poor statistics. The application of `central´ and `highest probability density´ confidence intervals is compared for the case of low count-rates. A method to determine realistic estimates of the confidence intervals for Poisson distributed variables affected with background, and their ratios, is given.
Print ISSN: 0033-8230
Volume: 91, 02/2003
Pages: 071