Article Abstract:
A technique for finding optimal sampling plans for Bayesian multiattribute acceptance sampling models is presented, in which inspections are assumed to be nondestructive and attributes are classified as scrappable or screenable (according to the corrective action necessary when a lot on a given attribute is rejected). The effects of interactions among attributes on the resulting optimal sampling plan indicates: that sampling plans for screenable attributes can be generated by solving a set of independent single-attribute models, that interactions of scrappable attributes on screenable attributes (and vice-versa) generate smaller sample sizes for screenable attributes than single-attribute plans, and that interactions among scrappable attributes generate either smaller sample sizes, lower acceptance probabilities, or both, in relation to single-attribute plans.
User Contributions:
Comment about this article or add new information about this topic:
Article Abstract:
The identification and control of misclassification error in a multistage screeningmodel is analyzed in the context of a hypertension screening program. Such a program identifies subjects as either normotensive or hypertensive, but the misclassification of one as the other is an error is a major but highly probable risk, with fatal results. Repetitive testing is so far the only current method of reducing such risks, but this results in higher costs and inconvenience that might reduce participation in such programs. Thus, a multiple-stage screening model that controls both maximum and mean misclassification error is proposed. Aside from minimizing misclassification risks at least to the level of repetitive testing programs, the proposed model also requires lower levels of subject participation.
User Contributions:
Comment about this article or add new information about this topic:
Article Abstract:
Bayesian models for single sampling inspection have become popular. The loss function they use is the linear cost function. However, there are times when risk aversion is important. A Bayesian model with risk aversion is constructed and then tested to see how it affected estimation errors.
User Contributions:
Comment about this article or add new information about this topic: