A note on test factors

Abstract

The calculation of test factors is reviewed. The distribution, of the population from which the test sample is taken is assumed to be Gaussian. Three cases are discussed, in which (i) there is no prior knowledge of the mean or standard deviation (if) there is no prior knowledge of the mean but the standard deviation is a given fraction of the mean (i.e., coefficient of variation known) (iii) there is no prior knowledge of the mean but the standard deviation is known. In each case an estimate is made of the average proportion of items under strength which go into service as a result of the continued application of a given test factor. The distributions of the statistics used in the solution of cases (i) and (iii) can be found from published Tables. The corresponding distributions for case (ii) for the appropriate ranges are given in this paper.