Hi Donald and Cyril,

Thanks for the feedback. Let me explain a little more what I currently have: I have a single-subjects model for each subject, where I input 10 conditions, and set up a contrast to compare condition 1 with condition 10 (ie 1 0 0 0 0 0 0 0 0 -1 with T-test)
Then I forward all subjects .con.nii's into a second-level, where I want to test the effect of a behavioural score, and how it changes between Trial1 and Trial10 (presumably using an F-test such as 1 0 0 0 0 0 0 0 0 -1). Based on some discussions on the SPM list, I thought I would look at it by entering a behavioural covariate (ie Trial10Performance - Trial1Performance). Hence, I thought I would have 1 value per subject.


Do you have any suggestions of how to include this covariate? I guess it's more of a time-varying covariate? Or a repeated measures with one behavioural value?

If you have other suggestions for design, that would be very much appreciated as well.

Thanks,
Joelle





On Tue, Aug 11, 2015 at 2:56 PM, MCLAREN, Donald <[log in to unmask]> wrote:
If you have repeated-measures, you should not be entering covariates as covariates aren't going to change the relationship between the actual measurements and if you have 1 covariate value per subject, it will be collinear with the subject columns. If there are only 2 repeated-measures, it would be easier to just create a difference image and then do a one-sample t-test with a covariate.

If you have a time-varying covariate, then you should be running a linear mixed effects model.

Best Regards, 
Donald McLaren, PhD


On Tue, Aug 11, 2015 at 1:57 PM, Joelle Zimmermann <[log in to unmask]> wrote:
Hi SPMers,

I'm setting up a second-level model, and I want to input a Covariate in the factorial design specification.

This is more of a technical question, what is the format in which I can input the Covariate vector? I'm not entirely clear on this based on the documentation in the GUI.

There it just says to "Enter the covariate values per subject (ie all for subject 1, then all for subject 2, etc)."

I have only 1 covariate value per subject, so can I just enter all these numbers in a list ie. [Sub1Covariate, Sub2Covariate, Sub3Covariate...] ?

Thanks,
Joelle