Department of Statistics Seminar
North Carolina State University

presents

Dr. Julian Besag

University of Washington

"Bayesian Analysis of Agricultural Field Experiments"

ABSTRACT

The main problem in the analysis of variety trials in the field concerns the low level or complete absence of replication among the test varieties, compounded by typically substantial spatial variation in the fertility over the experimental area, arising both naturally and because of management practice. In early generation unreplicated trials, it is common to also include plots of a standard variety distributed over the trial but otherwise there is no direct monitoring of fertility variation. Despite advances in spatial statistics, there was very strong resistance to the adoption of overtly spatial methods of analysis in field experiments until the 1990's and, despite the advantages of interpretation in ranking and selection, there has been little interest in Bayesian formulations. The talk will describe spatial methods for the analysis of field experiments in a Bayesian setting, implemented via Markov chain Monte Carlo methods. Several examples will be given. The talk will conclude with some remaining challenges arising in the formulations. Most of the material will be taken from the paper referenced below.

Besag, J. E. and Higdon, D. M. (1999). Bayesian analysis of agricultural field experiments (with discussion). Journal of the Royal Statistical Society B, 61, 691-746.

Friday, January, 18, 2002

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.