Use of additive models to represent trends in a barley field trial
Authors: HACKETT, C.A., REGLINSKI, T. and NEWTON, A.C.
Journal: Annals of Applied Biology
Volume: 127
Issue: 2
Pages: 391-403
eISSN: 1744-7348
ISSN: 0003-4746
DOI: 10.1111/j.1744-7348.1995.tb06683.x
Abstract:Even when a field experiment has been designed with care, subsequent examination of the plot values may reveal additional unforeseen trends. In this paper we examine data from a barley pathology field trial and show that additive models provide a flexible representation of environmental trends, in one or two dimensions. Such models smooth out noise in the observed data, rather than fit an equation specified in advance. This approach tends to increase the precision of treatment comparisons relative to a classical analysis of variance. We recommend the use of residual plots to explore experimental data for underlying trends, and additive models to display these trends and estimate treatment effects. Copyright © 1995, Wiley Blackwell. All rights reserved
Source: Scopus