Hi Gang,
Thanks for your input. In ran the analyses with the SPMG3 option. When I compare the coefficient from spmg3 (sub-brick 0) with the coefficient from the GAM method, the coefficients for spmg3 are substantially larger. Typically, the coefficients are about twice as high for spmg3. However, there are a few select voxels were the coefficient value is as high 30. Of course, the basic pattern is highly similar.
I'm curious if the coefficient from sub-brick 0 in spmg3 can be treated as a percentage change value as I would have done for the GAM-based coefficient.
Jatin