A simple bootstrapping procedure to validate the MERINOSELECT model for weaning weight

Abstract

Sampling distributions for the regression of progeny performance on sires' estimated breeding values (EBV) were estimated for weaning weight in the MERINOSELECT database. Mean regression coefficients from these distributions were used to compare the effects of data quality and the model used for evaluation. Results showed that better quality data with dam pedigree and date of birth recorded, and an evaluation model which included sire by flock-year interaction, improved the prediction of progeny performance from estimated breeding values. Sampling was conducted at three different strata, the most effective division being at the level of contemporary groups.