Multiple Attributes Sampling Acceptance Plans Involving Destructive Testing

Abstract

This article considers the following situation. A product unit is characterized by several different attributes and considered defective (i.e., overall defectiveness) if any attribute is defective. Whether or not a product unit is defective for a given attribute can only be determined by a destructive test, however—meaning that only one attribute can be tested for a product unit. Under the assumption of independence of the occurrence of defective attributes (P _i 's) and if the probability of overall defectiveness (P _o ) is small, then P _o can be approximated by Σ _t P _i . Furthermore, the convolution of independent binomial (n, P _i ) variates is approximately binomial (n, Σ _i , P _i ,) for P _o small. Therefore, we can define an acceptance plan (n, c) on P _o , sample n product units from each attribute, and accept it if the sum of the defectives is not greater than C. In essence, this article extends the single attribute sampling acceptance plan (SASAP) methodology to multiple attributes sampling acceptance plans involving destructive testing (MASAPDT).

KEY WORDS:

• Sampling for attributes