Weighted Inspection Sampling Methods Based on Winsorization

Abstract

Sampling inspection methods used in industrial quality control normally take the form of inspection-by-attributes or inspection-by-variables methods. Inspection-by-attributes sampling plans are noted for their robustness with respect to the distribution of the characteristic of interest. Inspection-by-variables sampling plans, on the other hand, are designed based on an assumed continuous distribution (usually a normal distribution) and, therefore, are not necessarily robust, as departures from this assumed distribution are encountered in practice but do permit relatively smaller sample sizes than would be required under an equivalent attributes sampling plan.

In this paper the author provides a new method for sampling inspection. The sample size levels and robustness of the new method lies in between the two classical inspection-by-variables and inspection-by-attributes sampling plans. The new method is designed and explained, and its equivalence to the classical methods is established. The sample size performance is thoroughly investigated and compared for the traditional and equivalent new methods. Their robustness is discussed at a preliminary level.

• attri-var sampling plans
• batch inspection
• equivalence of methods
• robustness

Additional information

Notes on contributors

Malik Beshir Malik

Malik Beshir Malik was born in Sudan. He received his bachelor's degree from the University of Khartoum. After graduating with honors with a degree in statistics, the university appointed him as a teaching assistant in 1976 and granted him a scholarship for graduate studies in the United Kingdom, where he received a doctorate in mathematics from the University of Essex, England. He returned to the University of Khartoum where he taught and became chair of department and later registrar and assistant dean. In 1994, he received the Fulbright senior visiting scholarship visiting Georgia State University. He taught at the Computer Learning Centers as lead instructor before he joined the University of Maryland Eastern Shore as an assistant professor in 1998. Later he became associate professor in the Department of Mathematics and Computer Science. He has written many papers in quality control and statistics and is currently preparing a handbook of statistical analysis.