Hi hopefully someone out there can help??
I am interested in a number of behavioural scores (and the brain areas
associated with these).
Some of the behavioural scores on tests in my dataset correlate
significantly with eachother (others do not).
Using VBM I have created a multiple regression model with a number of
variables (which contain the behavioural scores on various tests).
I was wondering, if I want to see which areas share significant
correlations, for example, on two of the tests, do I do a conjunction
analysis selecting the two tests of interest, or should I use inclusive
masking to examine this? I am not really sure of the difference, if
someone could clarify this and offer advice on which method would be
appropriate for me it would be appreciated.
A further question is that when I ran a conjunction analysis (selecting 2
of the tests - contrast 2 and 8) a new contrast was, as expected, created
in SPM. After this happened when I looked at the contrast however, it was
called:
022{T} Test1 (orth. w.r.t{8})
The first time I ran this conjunction analysis no significant areas were
revealed (at the chosen threshold). After the new contrast was created
(022{T} Test1 (orth. w.r.t{8})) and I ran this contrast again there were
significant areas (at the same threshold). What is going on here?
Why does SPM show no significant areas the first time and write this other
contrast which does show significant areas if I run it again?
Which one should I be interested in, the first time I run the conjunct
analysis or the analysis which is provided if I run the newly written
contrast again (or alternatively should I stick to inclusive masking)?
Looking at the design matrix for the new contrast that is written, the
bars which are created over the factors of interest change slightly
between the first time I run the conjunction and the second time (when
using the contrast that is automatically written).
Hopefully someone can help, I'm confused.
Thanks in advance.
Will
|