On Mon, Mar 12, 2012 at 11:07 PM, Annchen Knodt <[log in to unmask]> wrote:

Hello Everyone,

I am analyzing repeated measures data for the first time in SPM8, and I just wanted to verify that I am doing so properly. We are scanning subjects 3 times using the same task, and are interested in what areas show differential responses as a function of time (e.g. Are responses greater at time 1 than at time 2 and 3?).

In order to test this question, I am using the Second Level: Full Factorial option.

You should use the flexible factorial design and specify a subject factor in addition to the time factor.

I specify one factor with three levels (time, at visit one, two, and three). Under “Independence” I select “No” because it is a within-subject design. I then specify three cells, and enter the contrast images from my first level analyses for time 1 into the Level 1 cell, the con images from time 2 into the Level 2 cell, and the con images from time 3 into the Level 3 cell.

After design specification, and estimation, I create the following contrasts:
In order to test what areas show a main effect of task across all 3 times, I create an F contrast:

1 0 0
0 1 0
0 0 1

Once you add subjects, you will need to add zeros over the subject columns and the contrast would be:
1 -1 0
0 1 -1

Your contrast tests whether any time is significant, not if there are differences across time. The main effect of task/time is the difference across tasks/times.

In order to test which areas show a significantly different response at different times, I create an F contrast:
2 -1 -1
-1 2 -1
-1 -1 2

This is redundant to the main effect.

Am I analyzing the data correctly in order to test my question? Also, are the F contrasts I am creating appropriate for those questions? Any feedback you would be able to provide would be appreciated. Thanks in advance.

Annchen Knodt

~~~~~~~~~~~~
Annchen Knodt
Graduate Student
Computational Biology & Bioinformatics Program
Laboratory of Neurogenetics

Duke University