Journal of Official Statistics, Vol.5, No.4, 1989. pp. 365–374
Probability Estimation With Sample Compositing Techniques
Forest C. Garner, Martin A. Stapanian, Evangelos A. Yfantis, and Llewellyn R. Williams
Abstract:Sometimes the objective of an experiment is to estimate the proportion p of individuals that possess a characteristic, such as a parasite, a blood disease, or an antibody, and it is not necessary to identify those individuals. In such a case, analyzing composites of aliquots from k individual test portions may result in an estimator of p with substantially lower mean square error than the traditional estimator. The mean square error varies with k and p, we show a method for choosing the optimal value of k. Practical considerations for environmental scientists designing such an experiment are discussed. Substantial cost savings may result when a carefully planned sample compositing experiment is used.
Keywords:Sample compositing; group testing; probability estimation.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group