Hi Shary,
There's no real difference between covariates of interest and nuisance
ones (this has come up a few times on the mailing list); your
contrasts dictate which covariates you are interested in. I haven't
really used SnPM, but I think it allows you to specify a number of
covariates and then enter them.
There is a complication with adding covariates, which is whether to
orthogonalise them with respect to the rest of the design -- see this
paper:
http://dx.doi.org/10.1006/nimg.1999.0479
Best,
Ged.
Shahryar Rafi-Tari wrote:
> Dear all, I have 14 patients whose images (Gray matter/T1) I am going
> correlate wit h clinical measures. The only option I see in SnPM under
> SPM5 is single regression (multisubject or single subject) and SnPM does
> not offer any choice for covariates of no interest such as age, years of
> education etc. I need to remove the effects of age, years of education
> etc. out while correlate with regional diffusion with psychological
> measures. Have you designed such analysis before and used SnPM in SPM5?
> If so, how did you do it? If your answer is that there is no solution to
> this, it i s still a good answer, as everyone who is facing this problem
> will know. Thanks a lot in advance, Shary
|