Analysis of repeated measurements in Santa Ines sheep using mixed models

The objective of this work is to discuss and analyze longitudinal data in animal experimentation with Santa Ines sheep exposed to four different treatments through statistical methodologies. The treatments consisted of substitution levels of corn grain ground by palm oil pie (0, 15.30 and 45%), the analysis of the measurements was performed with the aid of the free software R and SAS. For the statistical analysis of the experiment with repeated measures in time, we used three statistical methodologies. (i) Univariate model, considered the plot of parcel subdivided in time, in which the plots are the treatments and the subplots being the time; (ii) Multivariate model, the structure of the covariances matrix is in the most general possible way, that is, variances and covariances may be different; (iii) The case in which it works with mixed models, where the structure of the covariances matrix can be modeled in the way that best represents the data. There was no significant effect of the inclusion of palm oil pie in the diet on the average weight of sheep animals. The univariate and multivariate methodologies are not suitable for the set of data in question, recommending the use of a mixed model for the experiment with repeated measures considering the dependence of observations obtained over time.