Comparison of consensus value methods used to compile on-axis gain measurement results

This paper compares methods for computing consensus values of data sets. Three methods are used to analyze multiple data sets. The data sets are, on-axis gain measurements that have a large range of values and uncertainties. The appropriate method is dependent on characteristics of the data sets such as, the number of data sets, the spread of the data set values and spread of the uncertainty values for each data set. One method computes a weighted mean where the weights are chosen as the inverses of the fractional error in the data values. A second method removes data sets that are determined to be outliers, then computes an unweighted mean. A third computes a simple unweighted mean. The results of this comparison show that the method chosen to compute a consensus value is fairly independent of the data sets.