> I'm wondering how this changes when I use a polort of 3?
> Do I still take p_0 coefficent of each run (subbricks 0, 4, 8
> and 12) and use the same expr?
If you are using the most recent version of AFNI, you don't need to worry about the other coefficients in your percent signal change calculation. The reason is that the Legendre polynomials are used in the current version of 3dDeconvolve, and they are well-centered in the sense that the area-under-the-curve is 0 for all polynomials except the constant as can be seen in the following figure.
> is all of this correct in the first place considering I use a
> wavered canonical?
It seems fine to me.
