In the Hoard's Dairyman Intel item "Genomic-only bulls dominate Hoard's List," I noticed the author fell into the trap when he said genomic evaluations provide twice the information provided by parent average alone. The direct statement was, "the genetic knowledge we gain is double pedigree parent average alone." That isn't correct.
Industry-wide, we have a major problem because people assume reliability means "accuracy" which isn't a direct correlation. I assume the author came to that conclusion because reliability is two times as high for genomic evaluations than it is for pedigree-based parent averages for numerous traits. That's isn't the complete story.
"Accuracy" is proportional to the confidence interval around a PTA. (Think the bell-shaped curve.) Therefore, "accuracy" is the proper measure of genetic knowledge. However, "reliability" isn't the same thing as "accuracy."
Both terms have clear scientific definitions. One must take the square root of reliability to obtain accuracy. When you take the square root of reliabilities, it becomes obvious that genetic knowledge is NOT double parent average alone.
Most people don't understand the science of reliability . . . versus accuracy . . . versus confidence interval. As the author alluded to, this may be one of the reasons sales of genomic bulls have overtaken progeny-tested bulls that have much more accurate, and thus stable, genetic evaluations. It will be among the many important topics discussed in greater detail at Advancing Dairy Cattle Genetics: Genomics and Beyond, February 17 to 19, 2014. For more information, go to www.ans.iastate.edu/events/dairygenomics.
- Les Hansen, Department of Animal Science, University of Minnesota