> Although 3dREMLfit allows to output both non-prewhitened and prewhitened residuals (-Rerrts and -Rwherr),
> it produces betas weights from the REML (prewhitened data) estimation only, right?
That's correct. However, the differences between the two sets of betas (with and without prewhitening) are usually negligible because of the unbiasedness property.
> I there any other way besides using 3dREMLfit to independently prewhiten the data? I want to avoid running the GLM analysis.
I'm not so sure what exactly you're looking for. Could you elaborate a little bit more?
Gang