What is needed to calculate EBVs and EBV accuracy?

1. Lots of data for a specific trait

a. Hundreds dogs of a particular breed.

2. Extended pedigrees entered in IWDR to tie dogs together

a. 5 generations is a general guideline

3. Data on all or most dogs in the population, not just the breeders.

a. Health clearance data on only certified dogs is not useful.
b. The dataset must contain all measures, the good and the bad.

4. Consistent measures –

a. Same measurement method or measurement methods that are correlated
b. Similar age brackets if age impacts expression of the trait

5. Accurate measures –

a. Skilled scorers with adequate inter-rater reliability
b. Measurement correctly done- test run correctly
c. Validated measurements

EBV accuracy is impacted by the amount of data on the individual and its relatives. The closer the relationship, the greater the impact.
In the chart below, the organization owning these studs obtains extended view and elbow quality measures on all puppies who have been raised in the puppy program. The accuracies are high for these traits. However, in most cases, only the breeders have PennHIP measures which provides fewer phenotype measures for the EBV software and therefore, lower accuracy. Keith has no progeny and his parents and other close relatives did not get PennHIP measures. The accuracy is so low that no EBV ranking for PennHIP displays.

Work is currently underway to seek statistical associations of locations on the genome and a particular trait. If significant genomic associations are found, these can be added to the data used to estimate of heritability and calculate the EBV to create an even more accurate EBV called a genomically enhanced EBV (gEBV).