Premium
Statistical Error Rates and Their Implications in Sugarcane Clone Trials
Author(s) -
Glaz B.,
Dean J. L.
Publication year - 1988
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj1988.00021962008000040003x
Subject(s) - type i and type ii errors , saccharum , statistics , mathematics , clone (java method) , crop , significant difference , set (abstract data type) , biology , agronomy , microbiology and biotechnology , computer science , dna , genetics , programming language
There is an estimated difference of 18 h ha −1 in the amount of time required to cut the two most widely grown sugarcane ( Saccharum spp.) clones in Florida, but this characteristic is neither analyzed nor considered in deciding if a clone should be released. The major objective of this study was to determine optimum type I and type II error rates ( α and β levels) at which to perform significance tests for this characteristic at our current sample size and error variance ( σ 2 ). The minimum detectable difference ( δ ) was set at 18 h ha −1 , the relative seriousness of type I versus type II errors was subjectively assessed, and the advantages and disadvantages of using various levels of α and β were then discussed. Analyses were performed for a group of 12 clones cultivated at five locations, from plant crop through second‐ratoon crop. It was determined most appropriate to use α = 0.10 so that β would approximate 0.04. The results indicate that at levels of α≤0.05, type II error rates are unacceptably high. For research projects in which type II errors have some negative consequences, researchers should estimate β at various a levels and set α at a level that appropriately balances type I and type II error rates.