Effect of Incomplete Records on Sire Evaluation

Abstract

First-lactation records of artificially sired Holstein daughters were examined to determine the effect of excluding incomplete records from sire evaluation. Results indicate that the average difference of incomplete records from stable-mate averages is different from herd level to herd level. The average difference is greater in high-producing herds than in lower-level herds. Incomplete records of daughters of high-level sires are closer to their stable-mate averages than are incomplete records of daughters of low-level sires. The effects of herd level and sire level are apparently additive in determining the average difference of incomplete records from their stable-mate records. Although the evidence indicates that a small differential bias would result in favor of low-level sires at the expense of high-level sires, if incomplete records are excluded from sire evaluation this bias is so small that it is unimportant, because of the approximately constant fraction of incomplete records of all first records over all sire levels and because of the small fraction of incomplete records (5-7%).