Abstract
Capability indices measure the proportion of process output which lies outside two specification limits. The index, C[p], is used to quantify the potential of a process to stay within the specification limits. The indices, C[pk] and C[pm], are used to quantify the actual performance of a process. For independent normal data, these indices have simple interpretations and are viable benchmarks for monitoring processes. However, when the process data has a non-normal distribution, the indices may provide a very distorted portrayal of the process. The presence of correlation in data used to estimate C[p] or C[pm] may result in highly inefficient estimators and unreliable confidence bounds. This dissertation contains new estimation procedures which adjust current estimators for correlated data. Several new indices are developed for processes having non-normal distributions. The indices and the estimation procedures for these indices depend on the type of deviation from normality. An additional goal of this paper is to put forth generalized functions for quantifying the quality of a process beyond measuring the proportion within two specification limits.
Bartlett, Randy John (1993). Measures of capability under nonstandard conditions. Texas A&M University. Texas A&M University. Libraries. Available electronically from
https : / /hdl .handle .net /1969 .1 /DISSERTATIONS -1476043.