COMPARISON AMONG PRESENCE-ABSENCE AND COUNTS SEQUENTIAL SAMPLING PLANS USING OVERDISPERSION DATA.

Dilermando PERECIN1

José Carlos BARBOSA[1]

§    ABSTRACT: Sampling plans for overdispersion data, as that to occur on counts of insects, pest management, weeds distributions per area etc. were compared .Two Presence-absence plans were used, the sequential probability ratio test based on binomial distribution (SPRTB) and the sequential test based on the confidence interval for the binomial distribution (STCIB). For overdispersion simulated data it was used negative binomial distribution. The negative binomial and Poisson sequential probability ratio test were used as count plans. For sampling plan’s evaluation, it was considered mainly the average sample number and the percentage of not accept H1, which takes to a decision "do not treat with pesticide, as an example". When the overdispersion is low, count plans were more efficient than presence-absence plans, but with high overdispersion, less efficient; is the sense that the average sample number is smallest. An approximately equivalence between SPRTB and STCIB for alfa=beta=0.10 and limits in the SPRTB all about 20% of the reference was empirically verified.

§    KEYWORDS: pest management; sequential sampling; contagious distributions; overdispersion

 



[1] Departamento de Ciências Exatas da Faculdade de Ciências Agrárias e Veterinárias - UNESP- 14870-000 - Jaboticabal - SP - Brasil. Bolsista do CNPq.