I don't really see any drawback to including motion parameters for everyone. You are not should not be removing any power from your signal of interest, you are only reducing noise by explaining signal variance to due to motion.

-Michael

On Thu, Apr 30, 2009 at 2:18 PM, Esther Fujiwara <[log in to unmask]> wrote:
How about dummy motion regressors for those subjects whose motion parameters I don't want to include (trying to keep down the amount of data to throw out ...)?

Michael T Rubens wrote:
You need to keep the number of covariates constant across subjects.

-Michael

On Thu, Apr 30, 2009 at 1:59 PM, Esther Fujiwara <[log in to unmask] <mailto:[log in to unmask]>> wrote:

   Is it possible to include motion parameters in first level analyses
   for some
   subjects (i.e., those moving around more) but not others if all
   subjects are
   later combined in a second level analysis? I suppose not, but I am
   not sure why.
   Thanks for any comments!
   Esther




--
Research Associate
Gazzaley Lab
Department of Neurology
University of California, San Francisco




--
Research Associate
Gazzaley Lab
Department of Neurology
University of California, San Francisco