Does model adequacy checking mean "Checking the normality assumption that $\epsilon_{ij}\sim \mathcal{N}(0,\sigma^2)$"? More specifically, does it mean checking that the residuals, $e_{ij}=y_{ij}−\hat{y}_{ij}$, are distributed in this way?
And is an independence check performed by showing that the mean square due to treatment and mean square due to error independently follow a chi-squared distribution according to Cochran's theorem?
Mean square due to treatmentandMean square due to errorindependently follow chi-square distribution according to Cochran theorem ? – Cynderella Aug 17 '13 at 00:55