Department of Statistical Science
Duke University
presents:
David Hidgon
Institute of Statistics & Decision Sciences
Duke University
"Spatial Lattice Models and Bayesian Analysis of
Agricultural
Field Trials"
Abstract: Spatial fertility patterns are usually a significant, and sometimes dominant, feature in data from agricultural field experiments. Though generally not observed directly, their effects must be taken into account. In contrast to the classical framework, there seems to be no alternative but to introduce spatial modeling of fertility patterns under a Bayesian formulation. The orderly layout of agricultural field trials are ideally suited for lattice models for the fertility patterns. A variety of Markov Random Field models will be covered for fertility adjustment in one and two dimensions. Extensions that allow for outliers and "jumps" in the fertility process will also be considered.
One of the primary goals in later stage crop experiments is to select best yielding varieties. The Bayesian framework is also quite useful for dealing with ranking and selection as well as combining information from multiple trials through hierachical models (spatial adjustment is crucial for combining trials). The talk will focus on two main applications: 1) a problematical RCB trial conducted in El Batan, Mexico, for which outliers and fertility jumps are evident. 2) a collection of trials conducted for official corn variety testing in North Carolina.
Friday, March 1, 1996
11:45 - 12:45
130 Sociology/Psychology Building Any questions concerning the seminar may be addressed to Cheryl McGhee @ (919) 684-8029, e-mail cheryl@stat.duke.edu, or finger seminar@stat.duke.edu.