REPEATABILITY AS AN INDICATOR OF STABILITY IN CONTEMPORARY COMPARISON SIRE EVALUATIONS

Repeatabilities of sire summaries by contemporary comparison were a measure of stability of sire evaluation. Modified contemporary procedures of sire evaluation for first lactation records only were used to compute predicted differences for all Holstein bulls having ten or more daughters calving in any calendar year, 1964 through 1973. A subset of 455 bulls having repeatabilities of last available (final) proofs equal to or exceeding 90% was used to examine relationships between first available (initial) predicted difference, initial repeatability, and final predicted difference. Sire evaluations with low repeatabilities were more variable with a slightly greater tendency to decline than evaluations with high repeatabilities. The average decline of sire evaluations with low repeatabilities was due to less than expected substantial increases in predicted difference rather than to more than expected substantial declines of predicted difference. Final evaluations increased 1.53 kg per point increase in initial repeatability. However, initial repeatability accounted for only 1% of the differences in final evaluations while initial predicted difference accounted for 76%. Analysis of semen prices showed a substantial cost for repeatability for bulls of high genetic merit for production and/or type. Cost-benefit ratios indicated that dairymen with intensive sire selection programs often will make most economical semen purchases at intermediate to low repeatabilities. © 1979, American Dairy Science Association. All rights reserved.