MONTE CARLO INVESTIGATIONS OF SMALL SAMPLE BRUCETON TESTS
- 17 February 1967
- report
- Published by Defense Technical Information Center (DTIC)
Abstract
Monte Carlo investigations of Bruceton tests (twenty-five and fifty trials) show the following characteristics: (1) a bias in the estimate of the standard deviation causes predictions of reliability or safety based on such tests to be too optimistic; (2) there are additional biases (in the parameters G and H) which will cause underestimates of the errors of the mean and standard deviation; (3) there is little correlation between the actual standard deviation and its estimate as obtained from the short Bruceton test; (4) the order in which the items of the sample are tested has a serious effect upon the estimates obtained; (5) a poor choice of starting level can give misleading results when the step is small.Keywords
This publication has 0 references indexed in Scilit: