[Gretl-users] Re: OLS regression with heteroscedasticy problem