[R] Anova - adjusted or sequential sums of squares?

[Peter Dalgaard]

"michael watson (IAH-C)" <michael.watson at bbsrc.ac.uk> writes:

> I guess the real problem is this:
> 
> As I have a different number of observations in each of the groups, the
> results *change* depending on which order I specify the factors in the
> model.  This unnerves me.  With a completely balanced design, this
> doesn't happen - the results are the same no matter which order I
> specify the factors.  
> 
> It's this reason that I have been given for using the so-called type III
> adjusted sums of squares...

...and that is completely wrong!

If there ever is a reason for using Type III SSDs, it should be that
the results do not really depend "very much" on the order. This is
conceivably the case in "nearly balanced" designs. (I.e. it can be
viewed as an attempt to regain the nice property of balanced designs
where you can read everything off of the ANOVA table.)

If the unbalance is severe, then results simply *are* dependent on
which other factors are in the model - the effect of weight diminishes
when height is added to a model, etc. It's a fact of life and you just
have to deal with it.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907