Estimation of POD across laboratories


Poster title: How to estimate the inter-laboratory variability of probability of detection (POD) in detecting GM grains

Conference: MARCO-FFTC Joint International Workshop on Benefits and Risks of Genetically Modified Food Crops in Asia. 8-10 October 2013 Tsukuba, Japan

Author: Kohji Yamamura, Junichi Mano

MARCO poster (Yamamura and Mano) (PDF 304KB)

The following program estimates the variability of the probability of detection (POD) among laboratories. Beta distribution is assumed for describing the inter-laboratory variability of POD. The global estimate of POD and the confidence interval are calculated. The range and standard deviation of inter-laboratory distribution of POD are also calculated.

R function to estimate the beta-binomial distribution (text file, 6KB)

The following program yields another estimate of the variability of POD among laboratories. Probit-normal distribution is assumed, instead of beta-distribution, for describing the inter-laboratory variability of POD. This method is identical to that proposed by Uhlig et al. (2013). The glmer function is used for calculating the Wald confidence interval.

R function to estimate the Probit-normal distribution (text file, 6KB)

Related topics
Yamamura, K., Mano, J., Shibaike, H., 2019. Optimal definition of the limit of detection (LOD) in detecting genetically modified grains from heterogeneous grain lots. Quality Technology & Quantitative Management 16:36—53   DOI: 10.1080/16843703.2017.1347992.

Optimal definition of the Limit of Detection (LOD)

 

English Page

NIAES_logo

 
Statistical Modeling Unit, NIAES