ANOVA ssType3

["Manoel Pacheco" <mpacheco@glec-oh.com>]

I wrote the following code to perform a split-split-plot ANOVA

frag.aov <- aov(deltat ~ habitat + dist + source + Error(plot/trans),
                         data=fragdat)

As the data is unbalanced, there is a need to use type III sum of squares. I
have tried two different ways with no satisfactory results:

ssType3(frag.aov) # results are generated but the degrees of freedom of
                                error terms are altered (wrong!)

anova.lm(frag.aov, ssType=3) # No error message and no ANOVA table!

I would appreciate suggestions on how to use type III sum of squares, while
maintaining the correct degrees of freedom for residuals.

    Thanks,
                        Manolo