Estimated Breeding Values (EBVs)

Statistics of the genetic evaluation

Genetic Parameters
Genetic parameters, including the additive genetic variance and the residual variance were estimated using the Restricted Maximum Likelihood (REML) procedure.

  • Additive Variance is variance due to the average effects (or additive effects) of the genes underlying the specific genetic trait.  Because parent-offspring resemblance depends upon the average effect of a single gene of interest being inherited, additive variance is thus responsible for parent-offspring alikeness.  We don't know what the genes and mutations causing hip and elbow dysplasia are yet.  We infer the genetic underpinnings from teh pedigree relationships.

  • Residual Variance are factors in any variance analysis that cannot be attributed to a precise cause.

  • REML is a specialized maximum likelihood estimation of the genetic output provided to you.

  • Heritability of a trait is measured by estimating the relative contributions of genetic and non-genetic differences to the total variation in a population.
    • Mathematically, heritability was defined as the proportion of additive variance over the total variance which is the sum of additive variance and residual variance.


Distribution of Breeding Value and Accuracy

  • The dogs with low breeding value (low OFA score means a better hip or elbow) and with higher accuracy (more related dogs measured, the higher the accuracy) are the most desirable for breeding purposes.

  • Low accuracy means that not many progeny were available to estimate the breeding value. Clearly, many of the dogs recorded in this web site will no longer be available for breeding as the OFA data base has records back 50 years.


Statistical Model
A two trait (hip and elbow OFA score) mixed linear model was employed to estimate variance components (additive genetic variance and residual variance) and to predict estimated breeding values for each dog in the pedigree.  The model is described here.

  • Accuracy of the breeding value was indicated by R-Square: R2=Square root (1-prediction error variance/genetic variance).