That approach and the design setup appear sensible enough to me. However, it is my understanding that the covariates in FSL still need to be centered (i.e. the average value of all values in one column in the covariates should be zero). So 1 -1 for gender is OK if you have equal males and females. The age column need to have the mean age subtracted from all values, which then gives a mean of zero.
Good luck!
Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Colm McGinnity
Sent: August-08-16 11:18 AM
To: [log in to unmask]
Subject: [FSL] GLM query - 3 factors plus covariate
Dear Experts,
I would be very grateful for advice regarding this GLM setup (below). I have seen similar topics addressed in the email list, but I am still a bit uncertain so would appreciate some help.
I have 3 factors (group - 3 levels), gender (two levels) and handedness (in this case, left or right, i.e. two levels). I also have a continuous covariate (age). I do not have repeated measures. I would like to create a design that allows me to compare between groups. I am particularly interested in directionality of the group effects, i.e. increases or decreases, whereas I am not particularly interested in the effects of gender, handedness, and age - I just wish to control for these.
This is what I suggest, EV1 - EV3 for each group, EV4 for gender, EV5 for handedness, EV6 for age, assuming 3 subjects per group:
EV1 EV2 EV3 EV4 EV5 EV6
1 0 0 1 -1 22
1 0 0 -1 1 33
1 0 0 1 -1 21
0 1 0 1 1 25
0 1 0 -1 1 46
0 1 0 1 1 34
0 0 1 -1 1 21
0 0 1 1 1 16
0 0 1 -1 1 53
My contrasts would be t-tests, for example Grp A > Grp B: [1 -1 0 0 0 0], Grp B < Grp C [0 -1 1 0 0 0], etc.
Is this correct and sensible? Is it ok to represent two levels of gender and handedness by -1 and +1 in the same EV?
I apologise for any misunderstandings and for this is rather basic question.
Thank you very much,
Best wishes,
Colm.
|