A method for computing the decision level for samples containing radioactivity in the presence of background

It is often necessary to make a decision whether or not a sample contains radioactivity in excess of background. The most common procedure is a deterministic approach that compares the result from each sample against a critical level, Lc. Originally Lc was derived from the case where each sample is paired with a blank or control that is known not to contain excess radioactivity. However, most analytic laboratories do not process a blank with each and every sample. Another approach computes Lc using a collection of measurements to form a "well defined" background. This paper presents a method for determining a decision level, L delta, that includes the uncertainties in both the mean and variance of background and how these combine to form uncertainties in estimating the tails of the background distribution. The values of L delta are greater than Lc for a given Type I error. It can be applied whenever the true distribution of background is either normal or log normal. The process is not restricted to counting statistics and is valid for identifying excess contamination of any type providing that unknown samples are processed identically to background samples.