Two-dimensional partitioning of yield variation: a critical note.
Marcin Kozak
marcin.kozak@omega.sggw.waw.plDepartment of Biometry Warsaw Agricultural University, Nowoursynowska 159, 02-776 Warsaw, Poland (Poland)
Abstract
The Two-Dimensional Partitioning (TDP) of Yield Variation is a method for studying a response variable (usually yield) as affected by successive traits contributing to it (first direction) and treatments (second direction). Many authors have found its usefulness, especially in plant breeding, but also in other agricultural and horticultural investigations. Since now, no disadvantages of the method have been pointed out. The objective of this paper is to discuss the statistical appropriateness of the TDP method. Two general problems are introduced, i.e., (1) employing sums of squares from ANOVA as factor effects, and (2) dealing with so-called cross-products, which cause that the TDP table is, actually, quite often very hard or even impossible to interpret. The author points out that inference based on TDP may be false and may lead to erroneous conclusions.
Keywords:
analysis of variance, Gram-Schmidt orthogonalization, sequential yield analysis, two-dimensional partitioningAuthors
Marcin Kozakmarcin.kozak@omega.sggw.waw.pl
Department of Biometry Warsaw Agricultural University, Nowoursynowska 159, 02-776 Warsaw, Poland Poland
Statistics
Abstract views: 36PDF downloads: 20
License
All articles published in electronic form under CC BY-SA 4.0, in open access, the full content of the licence is available at: https://creativecommons.org/licenses/by-sa/4.0/legalcode.pl .
