Approximate reliability of genetic evaluations under an animal model.

A method was developed for calculating approximate reliability for national systems of evaluation. The method combined the reliability of three information sources: parent average, animal's own records, and progeny records. This method provided good approximation to the actual values with minimal upward bias and was considerably better than the current national method of New Zealand genetic evaluation or Meyer's method for all accuracy measures. Our method had an average absolute bias of 0.006 compared with 0.026 and 0.035 for the current national method and Meyer's method, respectively. Our method was less computationally demanding than the current New Zealand method. One of the major advantages of the method is that it can be extended to accommodate more complex models by altering the selection index equations within the method. An example is given for which the method was extended to account for a genetic correlation other than unity between an incomplete lactation and a complete lactation yield.