ABSTRACT
We have been assuming, for the most part, that our multiple regression model relates a response variable Y to a set of regressors X1,…,Xm in a linear manner or
j = + +
= ∑β β ε0 1 , (17.1)
where ε is random disturbance term that, according to the strong classical assumptions, is N(0, σε), with σε constant. Here the fi rst two components involving the parameters β0, β1,…,βm appearing on the right side of Equation 17.1 constitute the parametric portion of the parametric specifi cation of the regression model.
