[MINC-users] mni.cortical.statistics help

[EJ Nikelski]

Hi,

   Whenever you run a regression using a categorical predictor, one of
the levels of that predictor is designated as the reference level; all
other groups are compared against that reference. So, I would suggest:

(1) use the "relevel" command to make sure that the reference group is
the one that you want.
(2) as we are using R and not a lesser stats package such as SPSS, I
suggest that you explicitly label your levels. See the "levels" and
"labels" arguments of the"factor" command.  This will make is easier
to see what your levels are on your output.
(3) you can, of course, run an analysis using one level as the
reference, relevel, and then run again using the new reference, if
that's useful to you.

Enjoy your weekend  ;)

-Jim




On Sat, Nov 8, 2008 at 10:21 PM, Mishkin Derakhshan <mishkind at gmail.com> wrote:
> Hi,
> I am using a simple glim file which contains
> filename,group
> but my group has three categories (group1,group2,group3).
>
> After running y~group here:
> vs <- mni.vertex.statistics(gf, 'y ~ group', vertex.table=dt)
>
> I look at the vertstats file and I have the following tstatistics:
> tstatistic.group1
> tstatistic.group2
> tstatistic.intercept
>
> What do these mean though? ie
> Does tstatistic.group1 =
> group1 vs (group2+group3) ?
> group1 vs group2 ?
> group1 vs group3?
>
> Is there a way to control the comparisons, or do i have to make
> separate glim files with a max of two groups in them?
>
> Ideally I would like to see
> group1 vs group 2
> group1 vs group 3
> group 1 vs group2+group3
>
> thanks,
> mishkin
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www2.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>



-- 
=================================
Jim Nikelski, Ph.D.
Postdoctoral Research Fellow
Bloomfield Centre for Research in Aging
Lady Davis Institute for Medical Research
Sir Mortimer B. Davis - Jewish General Hospital
McGill University
Tel: (514) 340-8222 x 2298
Fax: (514) 340-8295