Comparison of three sequential sampling plans using binomial counts to classify insect infestation with respect to decision threshold.
1994
Brewer M.J. | Legg D.E. | Kaltenbach J.E.
In insect sampling programs, decision making based on intensity often reduces sampling time because inspection of a sample unit is terminated once an infestation is found. Sampling effort can be further reduced by using sequential binomial classification plans in which the sample size necessary to make a decision, with respect to an economic threshold, is based on preselected error rates and information gathered as sampling units are sequentially inspected. Commonly used insect sampling plans are derived from regression-based confidence interval procedures or hypothesis testing probability ratio procedures. The plans differ in their construction and use of decision thresholds, but operationally, they function and are interpreted in a similar manner. Therefore, statistical behavior of these plans, as measured by average sample number, terminal decision, and operating characteristic curves, are useful to compare relative performance. Decision criteria of three plans with comparable decision thresholds were applied to simulated presence-absence data of known binomial means to construct these curves. The curves differed among plans, and the greatest differences were in the vicinity of the decision threshold(s). When comparing operating characteristic curves, these plans differed in their approximation to the operating characteristic of an idealized plan which functions without error. A common plan modification requires decisions to be made when a maximum sample size is reached (i.e., a truncation mark). Relative differences among plans changed when the truncation mark was varied. Based on these simulations, general guidelines may be considered when choosing among plans.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Cette notice bibliographique a été fournie par Wolters Kluwer
Découvrez la collection de ce fournisseur de données dans AGRIS