Design and evaluation of statistical control procedures: Applications of a computer 'quality control simulator' program

Abstract

A computer simulation program has been developed to aid in designing and evaluating statistical control procedures. The 'QC simulator' (quality control) program permits the user to study the effects of different factors on the performance of quality-control procedures. These factors may be properties of the analytical procedure, characteristics of the instrument system or conditions for the quality-control procedure. The performance of a control procedure is characterized by its possibility for rejection, as estimated at several different magnitudes of random and systematic error. These performance characteristics are presented graphically by power functions - plots of the probability for rejection vs the size of the analytical errors. The utility of this stimulation tool is illustrated by application to multi-rule single-value procedures, mean and range procedures, and a trend analysis procedure. Careful choice of control rules is necessary to minimize false rejections and to optimize error detection with multi-rule procedures. Control limits must be carefully calculated for optimum performance of mean and range procedures. The level of significance for testing control must be carefully selected for the trend analysis procedure.