Abstract
The practice of regressing dairy sire proofs for the numbers of daughters included and for the average correlation among those daughters has been gaining wide acceptance in recent years. However, the contributions of environmental and other extraneous correlations to the regression factors being used have been generally neglected. Justification for this has been based largely on the assumption that, when deviation records are used, the effects of these residual correlations in A.I. bull proofs are of no practical significance. In several studies with deviation records, the expected correlations of the separate proofs of bulls were computed on the assumption of zero residual correlations among the daughters represented in each proof. The wide discrepancies found between the expected and sample correlations raise some doubts as to the general validity of this assumption.
Keywords