A Criterion for the Construction of Optimal Neighbour Designs
Journal of the Royal Statistical Society: Series B (Statistical Methodology) · 1985 · Volume 47 · Issue 3
DOI: 10.1111/j.2517-6161.1985.tb01379.x
Abstract
The model of Williams (1986) for the neighbour analysis of field trials is investigated in order to derive a criterion for the comparison of neighbour designs and hence the construction of optimal neighbour designs. A two-stage procedure is given whereby an efficient resolvable incomplete block design can be used as a starting point for the construction of an efficient neighbour design. A definition of balance for neighbour designs is presented. A comparison with the method of Wilkinsonet al.(1983) is discussed.
Keywords
analysis of field trials, incomplete block designs, generalized lattice designs, alpha-designs, neighbour methods, linear variance model, average efficiency factor, optimal designs, balanced neighbour designs
Enable JavaScript for citation tools and additional paper interactions.