Normality test of parameters of in situ ruminal degradability curves.

Abstract

Starting from a carried out experiment at the Animal Science Institute, normal distribution of nonlinear regression parameter was analyzed using Orskov Model applied to in situ ruminal degradability curves. Data was from a complete randomized experiment with four treatments. The sample of parameters was obtained using bootstrap and 450 groups of parameters per treatment was obtained and tested with Shapiro-Wilks and Kruskal-Wallis tests and the Q-Q graphics. Hypothesis of normality was rejected for each treatment through all the tests. The distribution of the asymptotic parameter was the most near to normality. Results of graphical analysis show that transformation of parameter could lead to normality but it is possible to add a bias in the further analysis of the data. It is concluded that the parameters of the curves of non lineal regression for this type of experiments cannot be used in analysis of variance by not fulfilling the supposition of normality.