Hi Carlos,
Perhaps you could attach a snapshot of your design matrix and the
contrasts that you are testing?
cheers,
-MH
On Tue, 2011-09-06 at 23:48 +0100, Carlos Faraco wrote:
> Dear FSL community,
>
> I asked a question related to this a few weeks ago, but now have a more basic question.
>
> I am looking at a FA and RD differences between a clinical group and controls. We have acquired various neurocognitive and neuropsychological measures on these groups.
>
> One set of scores that I am interested in are RBANS scores; a neurospsych test that includes various domains, e.g., immediate memory, delayed memory, visuospatial constructional skills, etc.
>
> Most of the scores show significant differences between the groups, except for visuospatial scores. In regards to FA and RD alone, I see no differences between the groups.
>
> My question is what does it mean to add these as EVs/confounds/whatever you want to call them in a randomise analysis? In other words, if I were to find some significant results indicating FA differences in regards to immediate memory, what additional information am I gathering from this that could be useful clinically? Similarly, what if I found significant results in regards to visuospatial scores, which didn't show any differences previously?
>
> I know that at a basic level the randomise results are indicating there is a difference in the slope of the scores. I just would like an explanation of the bigger picture.
>
> Many thanks,
>
> Carlos
|