Hi,
On 19 Mar 2007, at 10:51, Benny Liberg wrote:
> Hi,
>
> I have a an experimental situation that consist of three stimuli with
> increasing intensity. I have two groups. I wouldn't like to model
> reciprocity and linear trends in both activations and deactivations. I
> wonder if this seem correct, and this is how I did it:
>
> FIRST-LEVEL (S1-3=stimuli with intensity 1-3)
>
> COPE 1: S1 1_0_0
> COPE 2: S2 0_1_0
> COPE 3: S3 0_0_1
> COPE 4: Positive linear effect -1_0_1
> COPE 5: Negative linear effect 1_0_-1
Looks good.
>
> I then model the each group seperately. I feed COPE 4 and 5 in/up
> to the
> second-level analysis to model both activation and deactivation
> where there
> is a linear effect:
>
> COPE 1: G1 +lin act 1_0
> COPE 2: G1 +lin deact -1_0
> COPE 3: G1 -lin act 0_1
> COPE 4: G1 -lin deact 0_-1
> COPE 5: G2 +lin act 1_0
> COPE 6: G2 +lin deact -1_0
> COPE 7: G2 -lin act 0_1
> COPE 8: G2 -lin deact 0_-1
I can't comment on that without seeing the 2nd level model.....hard
to interpret. We need to get the 2nd level analysis sorted before I
can interpret the rest of the email - though I'd be surprised if you
do indeed need 4 levels for this analysis.
Cheers, Steve.
> Now, if I assumed that the deactivation was greater the smaller the
> activation or the other way around (inverse trends), how would that
> matrix
> look? Suggestion: I then feed COPEs from the second level analyses
> in/up to
> a third level analysis:
>
> [COPE 2_3]COPE 1: G1 1_-1
> [COPE 1_4]COPE 2: G1 -1_1
> [COPE 6_7]COPE 3: G2 1_-1
> [COPE 5_8]COPE 4: G2 -1_1
>
> or that the amount of deactivation was "balanced" with that of
> activation...
>
> [COPE 2_3]COPE 5: G1 1_1
> [COPE 1_4]COPE 6: G1 1_1
> [COPE 6_7]COPE 7: G2 1_1
> [COPE 5_8]COPE 8: G2 1_1
>
> I then investigate group differences at the fourth level:
>
> [COPE 1_3=G1_G2]COPE 1: 1_-1
> [COPE 1_3=G1_G2]COPE 2: -1_1
> [COPE 2_4=G1_G2]COPE 3: 1_-1
> [COPE 2_4=G1_G2]COPE 4: -1_1
> [COPE 5_7=G1_G2]COPE 1: 1_-1
> [COPE 5_7=G1_G2]COPE 2: -1_1
> [COPE 6_8=G1_G2]COPE 3: 1_-1
> [COPE 6_8=G1_G2]COPE 4: -1_1
>
>
> Does this seem right? Thanks.
>
> Kind regards,
> Benny Liberg
>
>
>
>
>
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|