Hi Avihay,
I'm not sure if there is a best way but this is how I would do it
- per subject (1st level) model each session separately since there are
different days, and compute the contrast between conditions (since you
want to test this difference)
- do your 2nd level with 3 groups and 2 time points entering the
contrast defined above - differences between time points show the
interaction you are after
my choice is motivated here by removing the between condition variance
that is of no interest here ; one could argue to model all the data
together to remove the time valriance within subjects, but I personally
don't like the idea to pull together data from different sessions
(different runs is fine).
cyril
> Dear experts,
>
> I was wondering what is the best way to define my contrasts in 1st and
> 2nd levels for the following design and hypothesis:
>
> Design:
> 3 treatment groups (between)
> 2 time points pre and post treatment (within)
> 2 conditions (within)
>
> Hypothesis:
> Interaction between time and conditions would exist ( = be
> significant) in two groups but not the third.
>
> 1. In "contrasts manager", when I define the 1st level contrast for
> interaction between time and conditions
>
> [1 - 1 -1 1]
> (This is the specific direction of the interaction that I want to test
> for)
>
> Should I choose F contrast or t contrast? Is it possible to test for
> interaction effect with this single "t" contrast in SPM?
> I know the theoretical difference between the two tests, but since SPM
> lets me define the same contrasts for them, I'm not sure what to
> choose and why.
>
> 2. Let's say I choose the F contrast.
> Given that main effects are not part of my hypothesis, is it legit to
> define only the interaction contrast, without main effects contrasts?
> What is the statistical (or inferential) meaning of the difference
> between those two options?
>
> Thank you in advance,
> Avihay
--
Dr Cyril Pernet,
Senior Academic Fellow
Neuroimaging Sciences
Centre for Clinical Brain Sciences
Chancellor's Building, Room GU426D
The University of Edinburgh
49 Little France Crescent
Edinburgh BioQuarter EH16 4SB
[log in to unmask]
tel: +44 (0)131 465 9530
http://www.sbirc.ed.ac.uk/cyril
http://www.ed.ac.uk/edinburgh-imaging
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|