Nine populations, with five replications each, were simulated for average daily gain (GMD1) assuming normal distribution, average 100, and different group sizes as well as standard deviations. Each replication was divided into three different sets of contemporary (CG) and progeny groups. The first set (I) was formed by 1,000 contemporary groups, 100 observations each and 10 progenies/sire. The second set (II) included 2,500 CG, 40 observations and 4 progenies/sire. The third set (III) consisted of 5,000 CG, 20 observations and two progenies/sire. The number of records per group and sire and the number of sires per group varied within the same set. In each population, GMD1 was transformed by DIAS1=100/GMD1. The contribution of CG to sire breeding value predictions were calculated for GMD1 (Cx) and DIAS1 (Cy). The maximum and the average of DIAS1 effects on Cy absolute value were significant, but the R² were low (maximum of 16%). The minimum value of DIAS1 has not influenced Cy. The maximum and the minimum of GMD1 on Cx were significant, but the R² were very low (maximum 2%). The GMD1 average has not affected Cx. The effect of the minimum value of DIAS1 was important to determine the CG effect on breeding value of sires with progeny in the set. A very low value of GMD1 for an animal corresponds to a high value of DIAS1 and affects the values of the other animals and the CG average. This effect has an impact on the evaluation of their sires and will be an important source of error in the genetic evaluation of the herd. Thus, the replacement of GMD1 by DIAS1 as selection criteria for growth is not recommended because it will reduce the possibility of genetic gain.
average daily weight gain; heterogeneity of variance; selection criteria
