Articles
A NOVEL APPROACH TO THE PRESENTATION OF DATA FROM PROCESSING TOMATO VARIETY TRIALS
Article number
724_41
Pages
317 – 323
Language
English
Abstract
Data from processing tomato variety trials are often presented as a table of means for a number of different measurements.
Statistical significance is shown as a least significant difference derived from an analysis of variance.
While this format is instructive to those trained in statistical methodology, it often makes no sense to farmers who have little or no knowledge of statistics.
Furthermore, confusion often arises because of the volume and complexity of information presented in the one table.
A novel approach is described that utilises Residual Maximum Likelihood (REML) analysis, pairwise t-tests, an interactive computer program and a gambling analogy.
The model allows the user to compare the performance of two varieties in yield, holding ability, colour, soluble solids, solids yield and pH, and gives odds that a choice made is the correct one.
The user is also able to select the relative importance of each parameter as the basis for rating variety performance and calculating the related odds.
The model reduces the confusion that can arise from a complex table of information, by showing only two varieties at a time and the odds associated with their comparison.
It is envisaged that such a system will help growers, processors and seed companies to select varieties that warrant further assessment.
A simple training exercise to explain some basic statistical principles using a horse racing analogy is also described.
Statistical significance is shown as a least significant difference derived from an analysis of variance.
While this format is instructive to those trained in statistical methodology, it often makes no sense to farmers who have little or no knowledge of statistics.
Furthermore, confusion often arises because of the volume and complexity of information presented in the one table.
A novel approach is described that utilises Residual Maximum Likelihood (REML) analysis, pairwise t-tests, an interactive computer program and a gambling analogy.
The model allows the user to compare the performance of two varieties in yield, holding ability, colour, soluble solids, solids yield and pH, and gives odds that a choice made is the correct one.
The user is also able to select the relative importance of each parameter as the basis for rating variety performance and calculating the related odds.
The model reduces the confusion that can arise from a complex table of information, by showing only two varieties at a time and the odds associated with their comparison.
It is envisaged that such a system will help growers, processors and seed companies to select varieties that warrant further assessment.
A simple training exercise to explain some basic statistical principles using a horse racing analogy is also described.
Authors
G.R. Ashburner, B.M. Top, W.J. Ashcroft
Keywords
Lycopersicon esculentum, pairwise comparisons, statistical analysis, REML, technology transfer
Online Articles (45)