POTENTIAL BIASES IN PREDICTED TRANSMITTING ABILITIES OF FEMALES FROM PREFERENTIAL TREATMENT

Data were simulated according to the USDA animal model to determine potential biases in female PTA from preferential treatment. Ten scenarios for preferential treatment were investigated. Scenarios were defined according to three factors: number of records for the cows receiving preferential treatment, whether all or only second and later records received preferential treatment, and whether preferential treatment was added only to records of cows selected for preferential treatment or to selected cows and their relatives (dam or dam and maternal sibs). Three levels of preferential treatment were studied. Each was studied separately to allow straightforward determination of whether bias increased linearly as level of preferential treatment increased. For each scenario, bias increased linearly as level of preferential treatment increased, but magnitude of bias varied across scenarios. As a proportion of preferential treatment effect, bias in PTA ranged from .06 to .39. Affording preferential treatment to relatives increased bias more than increasing the number of lactations with preferential treatment.