DETERMINING THE BEST COVARIANCE STRUCTURE IN 33 FACTORIAL DESIGN
DOI:
https://doi.org/10.56892/bima.v2i01.59Keywords:
Factorial Design, Covariance Structures, First Order Auto-regressive, Compound Symmetry, Huynh-Feldt and Heterogeneous First Order Auto-regressive.Abstract
This research evaluated the performance of covariance structure in seed rates, row spacing
and varieties of bread wheat yield. 33 full factorial design method and mixed model methods
were used for the analysis. The four covariance structure used were compound symmetry
(CS), huynh-feldt (HF), first order auto-regressive (AR(1)) and heterogeneous first order
auto-regressive (ARH(1)). The goodness of fit criteria used to evaluate the performance of
covariance structures were Akaike Information Criterion (AIC) and Bayesian Information
Criterion (BIC). The data used composed of 270 observations for yield and were divided into
three factors and three levels. The factors were A, B and C and the levels are 1, 2 and 3 for
each factor. Analysis shows that the data satisfied all the assumptions. Based on the results in
this study, it was found that First order auto-regressive (AR(1)) was found to be the best
covariance structure for the data set.
