Factors affecting stability of contemporary comparison evaluations of sires for milk yield

Evaluations of sires for milk yield by contemporary comparison were used to determine factors associated with changes of evaluation over time. Changes in paired evaluations for 160 Holstein sires were final evaluations (Repeatability exceeding 90%) minus initial evaluations (Repeatability less than 70%). Multiple regression procedures were used to relate characteristics of initial sampling programs to changes in evaluations. Characteristics of initial evaluations associated with more stable evaluations included greater Repeatability for sires evaluated and for sires of contemporaries, higher producing, smaller sized herds with lower genetic merit of contemporaries, and less variation across herd-year-seasons in number of contemporary sires and producing ability of mates. Initial evaluations tending to increase over time were those from large herds of below average genetic merit (for contemporary sires) with greater than average variation in Repeatability of contemporary sires and less than average variance in number of contemporary sires and producing abilities of mates. Results suggested that most changes in evaluations of sires are from Mendelian sampling but that modest increases in stability of evaluations may be achieved through altered design of initial progeny testing programs.