|Van Vleck, Lloyd|
Submitted to: Journal of Animal Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/27/1997
Publication Date: N/A
Interpretive Summary: Models for statistical analyses commonly include effects of fixed and ran- dom factors of which examples for beef cattle are fixed breed of sire and random sire effects. Random effects, such as effects of genes transmitted by sires to progeny, have distributions showing variation. A fixed effect is said to be a constant, i.e., has no variability. This paper describes the consequences of ignoring random effects such as those associated with sires and dams in estimating differences in breeds of sire for birth, weaning, and yearling weight of crossbred beef cattle. The estimates of breed differences for records from the Meat Animal Research Center were not greatly different for analyses when effects of sires and dams were ignored and when those effects were included in the model. The confidence ranges, however, were greatly underestimated when the random effects of individual sires and dams were ignored. Because estimates of breed differences are used in comparing bulls across breeds, the errors in such comparisons are also underestimated when individual sire and dam effects are ignored in estimating breed differences.
Technical Abstract: Data on weights of F1 calves and on weaning weight of top-cross progeny from 13 breeds of sire and maternal grandsire breeds were analyzed. Three analyses were performed on each trait to obtain estimates and standard errors of breed effects needed to predicted across-breed EPD and accuracies. Model (R) for records F1 progeny contained fixed effects for birth year and date, sex, age and breed of dam, and breed of sire, and a random residual effect. The second and third analyses fit random effects for sires (RS) and for sires and dams (RSD), in addition to fixed and residual effects. In maternal analysis of top-cross progeny, model (Rm) contained fixed effects for cycle of experiment, age of dam, year of birth, sex, breeds of maternal grandam and grandsire, and breed of sire, and a random residual effect. In addition to these effects, the second and third analyses fit random effects for maternal grandsires (Rsm) and for maternal grandsires and daughters of maternal grandsires (RSDm). Estimates of bree of sire effects changed randomly for all models. Total variance increased in RSD relative to R or RS. Apparent standard errors of breed of sire comparisons were underestimated by R, compared to RS and RSD. Other standard errors were generally not affected. Variance components, breed effects, and standard errors followed similar patterns in Rm, Rsm, and RSDm as in R, RS, and RSD. Ignoring random variation due to sires and dams underestimates standard errors of breed of sire comparisons and underestimates prediction error variances.