Premium
SAS‐STABLE: Stability Analyses of Balanced and Unbalanced Data
Author(s) -
Magari Robert,
Kang Manjit S.
Publication year - 1997
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/agronj1997.00021962008900060013x
Subject(s) - stability (learning theory) , variance (accounting) , statistic , statistics , computer science , ibm , mathematics , materials science , accounting , machine learning , business , nanotechnology
Stability concept is often used to analyze genotype × environment interaction (G × E), by partitioning G × E variance into stability variance components for each genotype. A significant stability variance for a particular genotype indicates lack of stability. An existing computer program (STABLE) calculates the yield‐stability (YS i ) statistic for balanced data sets only. Therefore, we developed a program (SAS‐STABLE) that calculates stability variances for balanced as well as unbalanced data sets and is based on the restricted maximum likelihood method. Data may be stored in SAS files or other files that can be retrieved by SAS commands and statements used for data management. This program is for the Windows version of SAS for IBM‐PC compatible computers; a mainframe version is also available. Variance components are solved by iterating on some starting values. Prior knowledge of the values of variances would reduce number of iterations to reach convergence and save execution time. SAS‐STABLE can easily incorporate one or more covariates into the model and remove their effects from G × E variances for each genotype. Variances are tested based on the hypothesis H 0 : σ 2 = 0, assuming normal distribution.