Bootstrapping and Short-Run Process Capability Indices

摘要

The estimation of process capability indices (PCIs) in short-run situations has recently received attention in thernliterature. One approach that has been recommended is bootstrapping, which is a computationally-intensiverntechnique in which sampling with replacement from an original sample drawn from the process is performed manyrntimes. A potential issue is that each value of a bootstrap sample could be the same, causing the variance of such arnsample to be zero. The result would be an inestimable PCI. This paper investigates this issue and quantifies itsrnoccurrence for original samples with different numbers of identical values and sizes.