Statistical analysis of repeated measurements with modeling (co)variance structure

Repeated measures experiments have been used commonly in animal science for a long time، but only in recent years there have been big developments in methodological and computational issues that permit to analyze them effectively. Models and analyses of repeated measurements differ substantially from classical analyses. The many types of softwares are available in recent years to overcome disadvantages of the fitting models with repeated measurements. In this study repeated measurements from growth traits data of a Makoei sheep population were modeled by different variance-covariance structures. The models were compared by using Akaike’s Information Criterion (AIC)، AIC corrected for small sample sizes (AICC)، -2 Restricted log Likelihood (-2Res logL) and Bayesian Information Criterion (BIC). The “unstructured covariance (UN)” structure was chosen as the best structure for the data set. Furthermore، the most valid and the smallest standard errors were obtained using UN structure.