Analysis of discrepant data using a bootstrap procedure

We present in this paper a statistical procedure to deduce a best value and its standard deviation for a discrepant set of data. This method, rather than using the conventional average, uses a Bootstrap Method to estimate these quantities. The usual weighted average may be used to combine this estimate with additional data, thus avoiding a more complex calculation required by other methods, which would include all of the original experimental values. The current application of this procedure to the half-lives of 137Cs, 90Sr, and Tritium illustrates its usefulness.