We describe Bayesian spatial analysis of field experiments in which there is no replication of test varieties but checkplots for one or two standard varieties are included in the design to allow control for variation in fertility. Our spatial formulation provides unification between approaches based on geostatistics (kriging) and Markov random fields. The practical advantages of adopting the Bayesian paradigm in ranking and selection are emphasized. Examples are included.
See more of Analysis of Unreplicated Experiments
See more of Z06 Non-Traditional Experimental Design
See more of The ASA-CSSA-SSSA International Annual Meetings (November 6-10, 2005)