Abstract
Deviation records now constitute the basic data used by the USDA in com- piling sire summaries. Such a record consists of the difference between a mature- equivalent (ME) lactation record of a daughter and what is termed the regressed adjusted contemporary herd-mate average. The latter is derived from the simple herd-mate average by two adjustments designed to correct the herd-mate average for (a) unequal numbers of herd-mates included in the various deviation records in a proof and (b) differences in the average breeding values of the herds in which the bulls were used. Adjustment (a) is based on the ratio of the variance among records within herd-year-season groups to the component of variance between such groups. In the present data, this adjustment is approximately n/(n + 2) for nfilk and fat yields and n/(n + 4) for test, where n is the number of herd-mates included. The second adjustment (b) is the linear regression of a daughter's record on her contemporary adjusted herd-mate average, or approxiumtely .95 for nfilk and milk fat yields and .80 for test, in the present data. The variances among herd-year-seasons were only 21, 26, and 47% as large for milk, milk fat, and test, respectively, with deviation records as they were with nondeviated ME lactation records. With deviation records, the extra genetic gains that might be expected by including the records of the dams in sire proofs were only .7, 2.5, and 2.0% for milk, milk fat, and test, respectively. A major goal in sire evaluation is to obtain an accurate and unbiased estimate of the bull's breeding value. In practice, the conditions for an unbiased estinmte are rarely, if ever, com- pletely fulfilled. The problem is twofold: to identify the major sources of the extraneous variances that contribute to those biases and to reduce their effects where feasible. The herd is the main source, whereas the year-season of freshening is a nfinor source of environmental variance that affects a cow's record and thereby may contribute bias to bull proofs. In addi- tion, some genetic bias also may be introduced by genetic differences among the herds in which the records are made or, more specifically, by the mates of the sire in question being genet- ically better or worse than the mates of other sires. This paper concerns the role of devia- tion records, as presently computed by the USDA in proving bulls, in reducing these biases.

This publication has 7 references indexed in Scilit: