degrees of freedom for random factors in anovan
2 visualizzazioni (ultimi 30 giorni)
Mostra commenti meno recenti
"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?
0 Commenti
Risposta accettata
Shashank Prasanna
il 3 Apr 2014
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:
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.
Più risposte (0)
Vedere anche
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!