Hi

On 19 Apr 2010, at 20:13, Klara Mareckova wrote:

Hi,

could anybody please help me setting up my design for a voxelwise analysis of 2by2 ANOVA? I'm trying to follow the directions from http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html, but I'm not really sure if I'm setting up the EVs correctly.

My study has 40 subjects divided into 4 groups: on pill & menstruation, on pill & ovulation, free & menstruation, free & ovulation.

In the first level analysis, I had 5 conditions (EVs) where participants were were watching angry faces, angry hands, neutral faces, neutral hands or control condition. In the higher level analysis, I would like to explore the main effect of pill, main effect of cycle phase and their interaction.

Setting up the contrasts & F test seems pretty easy then:

pill                   100 0          on, off, off           

cycle               010 0          off, on, off

pill * cycle       001 0           off, off, on

What is more of a puzzle for me are the EVs:

1. Am I not working with my 5 conditions from 1st level analysis (angry, neutral,..) at all?

I don't quite understand - presumably at first level you will make a contrast of interest (e.g. angry vs neutral) and then feed that up for all sessions/subjects into the higher level modelling?   You could then repeat that for each first-level contrast of interest?

2. If so, would the EV set up look like this?

group    cycle     group    interaction      grandmean

1            1            1        1                          1

1            1            1        1                          1

1            1            1        1                          1

1            1            1        1                          1

1            1            1        1                          1

1            -1           1        -1                         1

1            -1           1         -1                        1

1            -1           1         -1                        1

1            -1           1         -1                        1

1            -1           1         -1                        1

2            1            -1        -1                        1

2            1            -1        -1                        1

2            1           -1         -1                        1

2            1            -1        -1                        1

2            1            -1        -1                        1

2            -1           -1        1                          1

2            -1           -1        1                          1

2            -1          -1         1                          1

2            -1          -1         1                          1

2            -1          -1         1                          1

Yes

3. With the set up above, I'm getting a warning: "design matrix uses different groups but these do not separable EVs for the different groups"

yes - this design is not "separable" so just set all the "group memberships" to 1.

Cheers.

Do you happen to have any suggestion what am I supposed to change to get the right design for my study?

Thanks a lot,

Klara 

---------------------------------------------------------------------------
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
---------------------------------------------------------------------------