degrees of freedom for random factors in anovan

"anovan" seems to give the same number of degrees of freedom to a factor whether it is declared as random or fixed.

I thought that random factors "cost" less degrees of freedom. At least this is what happens in regression with mixed effects.

Any clues? Is this normal behavior?

Answer 1

[Shashank Prasanna]

Laurie, anovan uses method of moments unlike in mixed-effects models which may use ML or REML. If you are interested in the latter, you may want to use:

http://www.mathworks.com/help/stats/fitlme.html

here the degrees of freedom will be equal to n – p, where n is the number of observations and p is ONLY the number of fixed effects, not random effects.

[laurie]

Ok, so this is a normal behavior! I thought I was doing something wrong with the model specification in anovan. Well, that's a shame. Thanks :-)