[Gretl-users] Normality, no autocorrelation, and no heteroscedasticity in Fixed-effect and Random-effects models