Avaliação da precisão experimental em ensaios de competição de cultivares de soja

Abstract

Data of 216 soybean [Glycine max (L.) Merrill] yield trials conducted in the State of Rio Grande do Sul during the 2003/04, 2004/05 and 2005/06 were used in order to evaluate the selective accuracy and other 15 statistics as a measurement of the degree of experimental precision of soybean yield trials. The minimum, maximum, average standard deviation and variation coefficient were obtained of 16 statistics in each experiment. The statistics selective accuracy, SA=(1-1/F)1/2, was correlated with all others. For each precision class (low, moderate, high and very high) according to the selective accuracy, the simple and relative frequency were obtained for each different categories (growth duration, year, significance among blocks and among genotypes) and general frequencies. According to these selective accuracy statistics, 76.4% of the soybean yield trials have a high and very high degree of precision whereas and only 6.9% have a low precision. The statistics selective accuracy is adequate to evaluate the experimental precision of soybean yield trials. It is not advisable to discard results of experiments for having a variation coefficient considered as too high.