The Use of Seemingly Unrelated Regression (SUR) to Predict the Carcass Composition of Lambs

The aim of this study was to develop and evaluate models for predicting the carcass composition of lambs. Forty male lambs of two different breeds were included in our analysis. The lambs were slaughtered and their hot carcass weight was obtained. After cooling for 24 hours, the subcutaneous fat thickness was measured between the 12th and 13th rib and the total breast bone tissue thickness was taken in the middle of the second sternebrae. The left side of all carcasses was dissected into five components and the proportions of lean meat, subcutaneous fat, intermuscular fat, kidney and knob channel fat, and bone plus remainder were otained. Our models for carcass composition were fitted using the SUR estimator which is novel in this area. The results were compared to OLS estimates and evaluated by several statistical measures. As the models are intended to predict carcass composition, we particularly focussed on the PRESS statistic, because it assesses the precision of the model in predicting carcass composition. Our results showed that the SUR estimator performed better in predicting LMP and IFP than the OLS estimator. Although objective carcass classification systems could be improved by using the SUR estimator, it has never been used before for predicting carcass composition.