On Mon, 20 Nov 2000, Joseph T. Devlin wrote:
> Following up from the recent discussion of conjunction analyses in SPM, I just
> wanted to post a short list of previous emails on this topic which may be
> helpful. I've broken it down into those relating to 1) SPM96, 2) SPM99b, and
> 3) SPM99. Note that this is not an exhaustive list.
>
> All URLs are preceded by www.mailbase.ac.uk/lists/spm/
> Finally, there has been some discussion at times about implementing conjunction
> analyses in RFX models. There didn't seem to be any kind of consensus so
> I'm wondering if there is a correct method for doing these or whether it isn't
> possible at all. Does anyone care to comment?
Yes, this is possible and there is a correct method, actually two methods.
See:
http://www.mailbase.ac.uk/lists/spm/2000-11/0005.html
for Karl Friston's suggestion for using the simple regression option in
"Basic Models" to carry out this analysis. This would seem to be the best
method offered so far, and it has the advantage of allowing corrected p
values to be calculated using GRF theory.
just for completeness, there is also:
http://www.mailbase.ac.uk/lists/spm/2000-11/0008.html
This is a solution using inclusive masking that gives the same result, but
one is limited to uncorrected p values. I've tried both on the same data
set and both methods lead to the same results.
regards,
Stephan
Stephan Hamann
Dept. of Psychology
Emory University
Atlanta, GA 30322
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|