Posted by Sean Drummond on April 02, 2002 at 15:03:51:
I have question about the GLT option in 3dDeconvolve, if I may.
My tasks have 3 levels plus a "resting baseline." In using 3dDeconvolve, I input 9 stim files: 1 for each task level (1's at the TR's with that level, 0's everywhere else) and 6 for motion parameters. I used stim_lags of 0-2 for each task level stim file. So, with -polort 1, that yields 17 parameters in the Full model.
I then set up 3 separate GLT's to look at each task level:
0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 Level1
0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 Level2
0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 Level3
I also set up a GLT to look, I think, at the linear trend with increasing task difficulty
0 0 1 1 1 2 2 2 3 3 3 0 0 0 0 0 0
And one to look at the overall task vs. BL
0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0
Problem is, the "Overall" GLT looks almost exactly the same as the "linear increase" GLT. This holds for 2 very different tasks. SO, my questions is:
Is there a better way to code the "Linear Increase" GLT? I did not use a Level1 = -1, Level2 = 0, Level3 = 1 contrast because I specifically wanted to take advantage of the fact that I have a middle level, rather than just ignore it. Should GLT contrast terms sum to zero?
e.g., Level1 = -2, Level2 = -1, Level3 = 3
e.g., Level1 = -3, Level 2 = 1, Level 3 = 2
As a secondary question: Once I have the GLT's for the individual levels, can I run a paired samples t-test to compare, for example, Level2-Level1? I imagine originally setting up a GLT to do so would have been best, but can I use a t-test to do it now?
Thank you very much for any input.
sean