Using regression coefficient as a stability parameter in plant breeding programs.

Expanding the regression coefficient as stability parameter (Finlay and Wilkinson 1963) requires an unbiased interpretation of the parameter. Information on the covariances among the genotypes of the population must be specific, particularly when the assumed relatedness of the genotypes appears questionable. Such a problem, however, is not expected when the covariances between the genotypes are either zero or equal and possibly non zero.