Hi,
To calculate broad sense heritability, we normally take genetic variance and divide it by total variances (for example, Genetic variance, GxE variance, error variance).
Proc Mixed (Method=Type3, or REML as default) is often used to get the variance estimate with a model following by a bunch of random terms.
However, if the analysis produce negative covariance estimate, what should we do with this negative number?
Thanks