Hi Mona,

You basically want to run the command like you will once you have
data. However the -nodata option is intended to evaluate your model,
not the data (it is intended to be used before data is collected, as
that is when you have the ability to alter your model). So using
-censor and motion parameters would usually not be done.

1. If you want to remove the first or last TRs from the analysis,
the -nfirst and -nlast options are the most convenient way to
go. But they have the same effect as using -censor. Neither
is necessary.

Use them in a way that will apply to all subjects. So if you
plan to always censor the first 4 TRs (as opposed to removing
them from the datasets), include the option.

2. You would probably not use the motion parameters, since they
apply to each subject differently.

3. Output goes to the screen as text, for later evaluation. The
sample @stim_analys script keeps track of it for finding the
'best' model.

[afni.nimh.nih.gov]

That ht03 directory is the one used for the class, and comes
from HowTo #3 (with changes).

4. The lower, the better, implying more statistical power. Since
those values are not absolute, the test is repeated many times
with different stimulus timing (in the script).

- rick