|Wright, Janice - Jan|
Submitted to: Interbull Annual Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 10/22/2000
Publication Date: N/A
Citation: Interpretive Summary: A strategy for multitrait genetic evaluation of US Jersey type traits that integrates an accounting for heterogeneous (co)variances was developed. This procedure improves the accuracy of genetic evaluations for type traits by accounting for the differences in variances across herds. It improves on procedures used for US yield traits by estimating the variance adjustment at the same time as the genetic values. This provides a more accurate variance estimate and therefore more accurate genetic estimates. By accounting for heterogeneous variances, bulls with daughters in primarily low variance herds should be more accurately ranked compared to bulls with daughters in primarily high variance herds.
Technical Abstract: A strategy for multitrait genetic evaluation of US Jersey type traits that integrates an accounting for heterogeneous (co)variances was developed. The proposed method allows joint estimation of variances and effects. The new model is theoretically better than the current one and should give less biased rankings of animals, especially for cows. The heterogeneous variance adjustment reduced standard deviations and linear regression coefficients of sampling variances on time substantially. Mendelian sampling variances for 1997 animals with and without adjustments were nearly identical. The HV adjustments were successful in stabilizing heterogeneous Mendelian-sampling variances,thereby reducing bias among animals evaluated at different times.