If you don't care about extending your results beyond your sample, then
an FFx is fine.
doug
SHAHIN NASR wrote:
> Do you suggest using random-effect model? Is there any problem with
> using a fix-effect model (other than the fact that by using this model
> we can not predict response of subject
Do you suggest using random-effect model? Is there any problem with using a
fix-effect model (other than the fact that by using this model we can not
predict response of subjects outside our population)?
On Wed, Oct 19, 2011 at 3:07 PM, Douglas N Greve
wrote:
>
> It should be roughly the sum of t
It should be roughly the sum of the dofs of the individual subjects. Why
are you using a fixed-effects model?
doug
SHAHIN NASR wrote:
>Thanks Doug. Just one related question. Should I also generate a
> new ffxdof.dat file for this map? I assumed that ffxdof depends on the
> number of subje
Thanks Doug. Just one related question. Should I also generate a new
ffxdof.dat file for this map? I assumed that ffxdof depends on the number of
subjects (session) and since number of subjects (sessions) is the same
between the two groups then I can use those values, generated by
isxconcat-sess
Hi Shahin, it is not as simple as doing a subtraction of the cesvar
files. What you are trying to get is the expected variance of your
difference between the ces files (as a variance, it must be positive).
To get this you need
cesvardiff = (cesvar1+cesvar2)/(2^2)
The 2^2 is the number of input
Hi,
I have generated two different functional connectivity maps for two
different ROI-based seeds. These maps are based on group-average of the same
subjects (generated by using separate isxconcat-sess commands ), and now I
want to see the difference map. To do so, I used mris_calc as below.
> m