On the Design of Experiments Under Spatial Correlation

Abstract

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.. Biometrika Trust is collaborating with JSTOR to digitize, preserve and extend access to Biometrika. SUMMARYThe implications for experimental design when the errors are spatially correlated are investigated. The first part of the paper considers the case when treatment effects are estimated by ordinary least-squares. If only treatment labels are randomized, then Aefficient designs are reasonably independent of the correlation structure, but this is not so for other efficiency criteria. If randomization of designs is considered within one of the classical design sets, then restricted randomizations may be chosen to reduce the variations in efficiency. The distributions of the usual sums of squares are also discussed. The second part considers the case when the analysis to be used is generalized least-squares using a known correlation matrix. Then for a particular correlation structure the efficiency orderings are relatively stable over different criteria, but there are important differences between long-range and short-range correlations whatever the criterion. The efficiency gain over ordinary least-squares is shown to be least for long-range structures. Predictions made from theoretical results on the torus (Martin, 1982) are shown to have some validity in the plane. In both parts numerical evaluations provide illustrations.

Parameter estimates for fractional autoregressive spatial processes

“…This led to the study of power law correlation functions by Whittle [25] and Besag [3] as an alternative to exponential decay. Martin [19] also indicates the importance of long-range correlation structures in agricultural field experiments.…”

Neighbour Analysis With Adjustment for Interplot Competition

“…The diagnostic test proposed by Ljung (1986) can be used for model selection by investigating the effect of model misspecification on estimation of treatment effects and variance parameters. By fitting models to real data, Martin (1986), Besag & Kempton (1986), Gleeson & Cullis (1987) and Lill et al (1988) suggest that for field trials, a low order AFUMA(p,d,q) model, with p = 0 or 1, d = 1 or 2 and q = 0 generally provides an adequate fit.…”

Constructing Pandiagonal Latin Squares from Linear Cellular Automaton on Elementary Abelian Groups

“…A Latin square is called pandiagonal if each entry appears only once in the main diagonal, the anti diagonal and all broken diagonals in both directions. These squares have several applications such as construction of pandiagonal magic squares [3], solutions of n-queens problem [4], and construction of statistical designs [12], where in this context they are known as Knut Vik designs, due to Vik [16]. A pandiagonal Latin square is cyclic if each row is a cyclic permutation of the first row and each column is a cyclic permutation of the first column.…”