Covariate in GLMFLEX

45 views
Skip to first unread message

soni...@gmail.com

unread,
Feb 20, 2019, 1:14:15 PM2/20/19
to Aaron's fMRI matlab tools.
Hi,

I was reading the previous posts about how to use covariates in the models but I got a little bit confused.
I have a model with 1 between group factor (group) and 2 within-group factors (condition1 and 2). I first built the model like this I2.Model = 'Group*Condition1*Condition2 random(SS|Condition1*Condition2)'.

Now I want to control for education level but I am not interested in the interaction effects between this covariate and the group, I just want to see if controlling for education changes the results of the first model. I did the second model like this I2.Model = 'Group*Condition1*Condition2+Education random(SS|Condition1*Condition2)'. Am I doing it right?

Thanks a lot!

Regards,

Sónia

soni...@gmail.com

unread,
Mar 19, 2019, 11:07:01 AM3/19/19
to Aaron's fMRI matlab tools.
Hi,

Can someone help me with this question please?

Thanks,

Sónia

Aaron Schultz

unread,
Mar 19, 2019, 11:18:02 AM3/19/19
to fmri_mat...@googlegroups.com
It's still not entirely clear what you are after with your model.
Education is presumably not a repeated measure, and does not vary
within subject. As a between subject factor, by definition it cannot
have any effect on the within subject factors. So the only place it
will change anything is in Group*X effects as education will modify
the Group variable.

You could ask a question about whether the difference between
condition A and condition B is related to education, but to do so you
would need to boil the data down to a fully between subjects model.
This would done in the standard SPM way of making specific contrasts
between specific conditions, and then analyzing those as a between
subjects test.

-Aaron
> --
> You received this message because you are subscribed to the Google Groups "Aaron's fMRI matlab tools." group.
> To unsubscribe from this group and stop receiving emails from it, send an email to fmri_matlab_to...@googlegroups.com.
> To post to this group, send email to fmri_mat...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/fmri_matlab_tools/d4bc34e5-8d15-426b-99d7-13df4399500d%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
Reply all
Reply to author
Forward
0 new messages