Fundamental procedures for the administration of an effective quality control program are given to enable clear understanding and application. Detection and control of determinate and indeterminate error are considered, with tables giving an application of the spiked sample procedure to the analysis of blood for lead content, an interlaboratory study of the determinate error in the dithizone procedure for the determination of lead in urine specimens, temperature correction for a Kitagawa carbon monoxide detector tube, and a typical graph showing skew, multimodes, and a narrow peak of a frequency distribution of lead in blood. With this foundation, the types of errors and meanings of the common terms used to define an accurate method are discussed as a basis for application of quality control to sampling and analysis. The theory, construction, applications and limitations of control charts are developed in sufficient depth to provide practical solutions to actual quality control problems. Additional statistical approaches are included to support those systems which may require further refinement of precision and accuracy to evaluate and control sampling and analysis reliability. Specific approaches discussed include rejection of questionable results, regression analysis, graphic analysis for correlations, chi square test, analysis of variance, and Youden's graphical technique. Finally, collaborative testing projects and intralaboratory quality control programs designed to improve and test the integrity of the laboratory's performance are discussed.
The Industrial Environment - Its Evaluation and Control, NIOSH, Cincinnati, Ohio, Chapter 22