The objective of this work was to evaluate the convenience of defining the number of multiplicative components of additive main effect and multiplicative interaction models (AMMI) in genotype x enviroment interaction experiments in cotton with imputed or unbalanced data. A simulation study was carried out based on a matrix of real seed-cotton productivity data obtained in trials with genotype x environment interaction carried out with 15 genotypes at 27 locations in Brazil. The simulation was made with random withdrawals of 10, 20 and 30% of the data. The optimal number of multiplicative components for the AMMI model was determined using the Cornelius test and the likelihood ratio test onto the matrix completed by imputation. A correction based on the data missing in the Cornelius procedure was proposed for testing the hypothesis when the analysis is made from averages and the repetitions are not available. For data imputation, the methods considered used robust submodels, alternating least squares and multiple imputation. For analysis of unbalanced experiments, it is advisable to choose the number of multiplicative components of the AMMI model only from the observed information and to make the classical estimation of parameters based on the matrices completed by imputation.
Gossypium hirsutum; unbalanced data; data imputation; AMMI models
