Comparison of Procedures to Convert Sire Evaluations Between Countries

Open Access
Publisher Website
Google Scholar
Abstract
Four methods for determining intercepts and regression coefficients of formulas to convert sire evaluations from one country to another were applied to data from Canada to estimate US PD. Four data sets were used: information from bulls with 1) Repeatabilities of at least 55%, 2) Repeatabilities of at least 75%, 3) Repeatabilities of at least 75% and initially sampled in the exporting country, and 4) Repeatabilities of at least 75% and young enough so that daughters with records used in calculating evaluations could not have resulted from decisions based on an evaluation in the other country. Use of assumed scale relationships of 53, 1.96, and 1.68 kg/breed class average unit for milk, fat, and protein was found inappropriate, especially for fat. Ordinary least squares method produced reasonable intercepts but small regression coefficients. Methods attributed to Goddard and Wilmink and recommended by the International Bull Evaluation Service generally produced similar equations and were of similar accuracy in predicting PD from independent samples. The Goddard method is recommended on an intuitive basis. Results were affected more by choice of data than by choice of method. The data set judged most appropriate contained information for bulls with Repeatabilities of at least 75% that were initially sampled in Canada, the exporting country.

This publication has 4 references indexed in Scilit: