We consider regression of the Γ^j coefficients on the γ^j coefficients where the intercept is not constrained to be zero. We fit the linear model: (4)Γ^j=β0E+βEγ^j. (We draw attention to the slight oddity in notation: the Γ^jand γ^j association estimates are the data in this model, and β0E and βE are the coefficients in the regression model; estimates of these coefficients are also denoted by hats –β^0E and β^E.)
