I am undertaking a nested ANOVA, where there are 3 levels of the nested
factor at each level of the main factor. For each level of the nested
factor, there are anything between 1 and 4 measurements.
Is it worth testing for the usual assumptions on the nested factor as i
expect the low replication will result in very low power tests for e.g.
normality and equality of variances.
Would I just be better of testing for assumptions on the main factor only
and then looking at residual plots ffrom the model fit.
Thanks
Alan Sharpe
|