Statistical process control methodologies for quality improvement

A study conducted by the Ford Motor Company in late 1983 found only 50% of Ford suppliers' processes capable of meeting requirements. Common sense suggests that a process incapable of meeting requirements should not be used in the long run. Processes incapable of meeting requirements result in resources being allocated to the identification and repair/replacement of non-conforming output. Some aspects of process capability and its measure are examined.; Various properties of the process capability index are examined. The stochastic nature of the traditional estimator is stressed and analytical tools that promote stochastic interpretations and warnings are presented and discussed. The robustness of the traditional estimator with respect to departures from normality and a general procedure designed to detect departures from distributional assumptions are presented. As well, a Bayesian technique that alleviates some of the problems encountered in drawing stochastic inferences from the sampling results is suggested.; A measure of process capability is proposed that incorporates some of the new philosophies arising in quality control methodology. The new measure takes into consideration proximity to the target value. Some properties associated with a proposed estimator are presented and comparisons drawn among other competing measures. A multivariate analogue is presented and discussed as well.; Finally a graphical procedure for monitoring a process is presented. The procedure provides an alternative to the boxplot style of simultaneous control charting now being suggested in the literature and the traditional x and s control chart. It provides information regarding the process' proximity to the target value as well as the variability for both the univariate and multivariate cases.