STABILITY OF EVALUATIONS OF BULLS SAMPLED BY ARTIFICIAL-INSEMINATION AND OTHER ORGANIZATIONS

Changes in evaluations during first- and second-crop periods were computed from up to 8 Modified Contemporary Comparison evaluations during each period for 1232 AI- and 217 non-AI-sampled Holstein bulls; repeatability at last Modified Contemporary Comparison evaluation was greater-than-or-equal-to .90. Evaluations were numbered consecutively within period. Effect of evaluation number within testing period on the bull's milk or fat evaluation was estimated with a model, including fixed effect of bull in addition to evaluation number. For subsets of AI- and non-AI-sampled bulls born from 1970 to 1975 and from 1976 to 1981, evaluations of both types of bulls increased from the initial first-crop evaluation and then remained fairly constant during die remainder of the first-crop period. With the inclusion of second-crop daughters, evaluations decreased significantly more for non-AI- than for AI-sampled bulls born from 1970 to 1975 (milk, 108 vs. 28 kg; fat, 3.5 vs. 1 kg). The reduction in evaluations was larger for non-AI-sampled bulls born after 1976 (milk, 184 kg; fat, 7 kg) but was unchanged for AI-sampled bulls. Last second-crop evaluation minus the second to last first-crop evaluation (difference 1) was calculated. Expected standard deviation of change for each difference of each bull was calculated and used to stratify bulls into eight change classes. Percentages of bulls in each class were compared. Relatively more non-AI- than AI-sampled bulls had first-crop evaluations that decreased more than could be explained by chance alone. Results indicated that evaluations of non-AI-sampled bulls were less stable than evaluations of AI-sampled bulls and that stability of evaluations of non-AI-sampled bulls has diminished over time. It appears that daughters in the first-crop of many non-AI-sampled bulls have received some kind of preferential treatment.