Hi Katie -
I'm afraid, if all of your first level models are different, then I can't
think of a clever way to calculate the average %signal change, without
simply computing it for each run and averaging.
If you want to do stats on this data then, you're right: it is important
to be careful with your degrees of freedom etc. You should first average
each of the pairs of data points from the same subjects, leaving you with
13 data points, then you can do a random effects t-test on these 13
data points with 12 dof.
hope this is useful
cheers
T
-------------------------------------------------------------------------------
Tim Behrens
Centre for Functional MRI of the Brain
The John Radcliffe Hospital
Headley Way Oxford OX3 9DU
Oxford University
Work 01865 222782
Mobile 07980 884537
-------------------------------------------------------------------------------
On Mon, 10 May 2004, Katie Karlsgodt wrote:
> Hi,
> I have 13 subjects who each have data on 2 runs of a task. I've combined
> the runs for each subject in a second level analysis, and then combined the
> subjects into a group in a third level analysis. Now I would like to get
> percent signal change data on a couple ROIs, and am not sure if there is a
> way to use the output of the higher level analyses in Featquery to get that.
> I've read some things on here about using the higher level output, but don't
> know what to do about the multiple runs or the percent signal change.
> My main concerns are: 1. being sure that each subjects' data is
> combined (so that it doesn't calculate it as if there were 26 subjects); and
> 2. how to get around the fact that the height of the wave forms for each
> run is different in design.mat- meaning that if I want percent signal change
> each run has to have the Featquery-calculated percent signal change
> corrected for separately (the design is event-related and the heights are
> not 1). Right now I'm getting the output from each run, correcting for the
> model height and then averaging the resulting numbers in excel. I'm mainly
> wondering if there is some way to make use of the higher level stats to get
> Featquery to do this for me, or if this is the only way. Any advice would be
> appreciated!
>
> thanks,
> Katie
>
> ___________________________________
> Katie Karlsgodt
> Dept of Psychology/Cognitive Neuroscience
> University of California, Los Angeles
>
> [log in to unmask]
> (310) 794-9673
>
|