Poder estatístico e coeficiente de variação em experimentos com bovinos de corte

Abstract

This study was conducted to estimate power of tests in Analysis of Variance (ANOVA) in beef cattle experiments, determine sample size and classify the coefficient of variation. Data was collected from thesis and dissertations of the Program of Post-Graduation in Animal Science of Federal University of Santa Maria (PPGZ-UFSM) among the years of 1991 to 2012, in beef cattle production area using ANOVA. Power was estimated by assumption of non central F distribution to the alternative hypothesis in ANOVA. The number of replications for treatments was estimated by power of 80% and interval of [0,4 ; 2,0] to effect size. Classification of coefficient of variation was based on proposed by Garcia (1989) using mean and standard deviation, and Costa et al. (2002) using median and pseudo-sigma. Power of tests shows two frequency peaks of experiments with low and high power. The recommended average number of replication per treatments was among 7 and 10 replications. Classification of coefficient of variation was proposed by low, medium and high, differently from literature that considers the intervals low, medium, high and very high.