Consider the linear model yi = 0 1xi1 pxip ei for i = 1,...,n where Ee = 0 and Cov(e) = 2In. Let x j = ni=1 xij/n denote the sample mean of the jth explanatory variable for j = 1,...,p