Methods for classifying coefficients of variation in experimentation with poultrys

Abstract

The aim of this work was to evaluate the coefficient of variation (CV) and to establish classification ranges for the main variables evaluated in experimentation with poultrys through different methods. The CV data of different response variables observed in poultry articles (broilers, laying hens, roosters and quails) published in five national journals were tabulated in a Microsoft Excel 2010® worksheet. Afterwards, they were subjected to normality evaluation through the Shapiro-Wilk method, and later the following data of the descriptive statistics were utilized: maximum value, minimum value, range, mean, median, standard deviation, midquartile, first quartile, third quartile, interquartile range, and pseudo sigma. According to the values obtained in the descriptive statistics of the CVs of each variable, classification ranges were elaborated for these values, considering six different forms of classification, adapted from the consulted literature. The results indicated that the classification ranges established for CVs in experiments with poultrys are different from those recommended in statistical textbooks. It is possible to utilize midquartile, pseudo sigma, mean and standard deviation through different methods in order to classify the CVs, but the use of the standard deviation promotes incoherent classifications in certain variables, and promotes increase in the amplitude between the lowest and the highest CV value obtained for the classification ranges.