Design and analysis of sampling plans to estimate aflatoxin concentrations in shelled peanuts

Methodology for use in the design and evaluation of sampling plans to estimate aflatoxin concentrations in lots of shelled peanuts is presented. Use of the operating characteristic curve for comparing and evaluating processor and consumer risks related to various sampling plans and application of the negative binomial distribution to estimate probabilities associated with sampling lots of shelled peanuts for aflatoxin concentration are discussed. Operating characteristic curves are developed for two different single-sample plans, an attribute multiple sample plan, and the plan presently used by the peanut industry to estimate aflatoxin concentrations in commercial lots of shelled peanuts. An estimated prior distribution of lots according to aflatoxin concentration is used to predict, among others, such values as the per cent of all lots tested that will be accepted by the sampling plans and the average aflatoxin concentration in the accepted lots. All four of the sampling plans described in the paper are compared on the basis of values such as these. Other factors to be considered in the critical evaluation and selection of sampling plans for estimating aflatoxin concentrations in commercial lots of shelled peanuts are discussed. Paper number 3197 of the Journal of Series of the North Carolina State University Agricultural Experiment Station, Raleigh, N.C.