scientific article; zbMATH DE number 1107630
There are advantages in eliminating the need for voluminous tables of the beta distribution function in implementing Form 2 acceptance sampling by variables for unknown process standard deviation. National and international standards on variables sampling would then be more compact and offer greater flexibility. In this paper, some previously proposed approximations to the beta distribution function have been improved upon but still found to be insufficiently accurate for this purpose. On the other hand, the new method using \(u_{10}\) developed here turned out to be highly accurate, guaranteeing an error of not more than 0.0004 for sample size 5, 0.0002 for sample size 6, and 0.0001 for sample size 8 or more. For most purposes, this will be well within the accuracy of measurement of the sample values. For sample sizes 3 and 4 the respective exact beta distribution functions \((2/\pi)\sin^{-1}\sqrt x\) and \(x\) can be used. The new method is also flexible, as the user can just as easily estimate the process fraction nonconforming for non-preferred sample sizes. It makes use of normal distribution tables, which are needed anyway for implementation of acceptance sampling by variables when the process standard deviation is known. The penalty in achieving these levels of evaluation accuracy is that several calculation steps are necessary after evaluating the sample mean and the sample standard deviation. For the entire collection see [Zbl 0867.00026].
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4373565)
