On Fri, 3 Apr 2009 14:00:02 +0200, Laura Menenti
<[log in to unmask]> wrote:
>Dear Radek (and Thomas and Natasha),
>
>You asked whether you should specify the subject factor when testing for
>between-group main effects in a flexible factorial design. As you correctly
>noted, the between-subject variance is crucial for assessing the
>significance of between-group effects. Specifying the subject factor removes
>just this variance from the error term.
>
>Therefore, you should *not* specify the subject factor when testing for
>between-group effects. As we understand from your email, Radek, you were
>testing for the main effect of group.
<snip>
You're absolutely right that specifying the effect of subject will remove the
effect of group if you're not careful.
However, in textbook treatments of ANOVA, this isn't exactly true. You can
remove the subject effect after first looking at effect of group. In terms of
breaking down variance, the intersubject variance is split between group
differences and differences between subjects _within groups_.
In the context of SPM, the monograph by Gitelman and Glascher claims you
can specify the effect of subject and still consider the effect of group. (I'm
referring to "Contrast weights in flexible factorial design with multiple groups of
subjects," available as an attachment somewhere in these archives. If I can
find it I'll post the link.) I haven't checked their math, but I assume they're
correct.
|