Continuing a thread concerning the size of matrix needed for
a particular ANOVA, David Airey <[email protected]>
says:
> <cut>
>
>> When you compare the numbers from doing this to the degrees of
>> freedom for each of the terms, it becomes clear real quickly why
>> they call it the "overparameterized ANOVA model".
>
> I keep hearing this term, but I don't get the importance. A bad thing?
Not a bad thing in and of itself. The comparison is most often
stated in terms of the "cell means ANOVA model" versus the
"overparamterized ANOVA model". Both can provide answers to the
same questions. The overparametrized model is the standard
computational approach for ANOVA. It automatically provides F
tests for "terms" in the model. getting various single degree of
freedom tests in a moderate to complicated overparameterized
ANOVA is not so automatic or natural, however -- see the "Testing
coefficients" section of [R] anova to see what I mean.
