On the estimation of the size of a subgroup in industrial production: The test gate method

AbstractEven in high quality manufacturing, some units may exhibit deviations. Although such units are not actually nonconforming to usage requirements when sent to the field, the deviations may lead to a reduced lifetime. It is therefore imperative to estimate the number of units bearing such deviations for risk assessment. Standard sampling methods are inefficient for this purpose. As an efficient alternative, we suggest the test gate method. The method is demonstrated on two practical examples and we provide an empirical model with stochastic extensions. We discuss the test gate method and customary alternatives, in particular control sampling.