Item reliability based on a paucity of item failures
Alan J. Gross, Philip F. Rust · Communication in Statistics- Theory and Methods · 1987
When item reliability is high, lot acceptance sampling is likely to uncover no defective units. Nevertheless, the reliability lower confidence limits are often disappointing. The use of a Bayesian procedure with certain beta priors produces tighter confidence limits. The case where one observed defective is permitted in accepting a lot is also examined. Numerical results for both cases are presented. In addition, the impact of item misclassification is addressed.