Integrating Human Factors Models into Statistical Quality Control

Recent progress in the statistical quality control field has led to the design of sampling plans which do not assume perfect inspection. Simple methods now exist for analyzing the effect of inspector error on the operating characteristic (OC) curve of a plan and further for redesigning the plan so that a predetermined OC curve is obtained. However, the usual assumption made about human inspection error is that it is constant. Many studies show that Type 1 and Type 2 inspector errors change systematically with many variables such as input quality, complexity of item inspected, type of fault, standards, individual differences, etc. This paper develops a methodology for including an explicit human inspector model into the sampling plan design. A particular model integrating visual search and decision making (proposed earlier by the author) is used to demonstrate the feasibility of including explicit human inspector data in the design process. The applications of this model to single and double sampling plans are discussed, together with evidence for the validity of the model under laboratory and field conditions.