O.K., I'm coming back to this with additional info and following no prior
responses to see if I'm on the right track or chasing dragons.

Three questions:

1.) Is it even possible to interpret/disentangle a second-level ANOVA of IBS
con images for a first-level, condition vs. condition contrast? Or, is the
ANOVA IBS method just relegated to looking at condition vs. zero baseline?

2.) Assuming that it is possible to enter canonical, time, and dispersion
derivatives for first-level, condition vs. condition con images, what are
the time and dispersion columns in the (1 0 0, 0 1 0, 0 0 1) main effect
telling you?

3.) If number 2 above is not possible, and it is only possible to capture
signal using the ANOVA IBS method for condition vs. zero baseline, then
would the following masking scheme work to capture +/- canonical activity
while preserving the time and dispersion components:

(1 0 0, 0 1 0, 0 0 1) F main effect exclusion masked by (-1 0 0) t to yield
activity associated with positive canonical + bi-directional derivatives

(1 0 0, 0 1 0, 0 0 1) ..................................(1 0 0) t
........................ negative canonical + bi-directional derivatives.

Or, maybe the above as inclusion masks?

Jeff


-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Jeff Browndyke, Ph.D.
Sent: Saturday, February 24, 2007 7:12 PM
To: [log in to unmask]
Subject: [SPM] Masking for +/- effects when using ANOVA of IBS

As is typical form of an old computer geek, I'm puzzling over the use of the
IBS within an ANOVA. One of the studies I'm working with has a condition
that would benefit from the capturing of signal variance 1 sec post-stimulus
onset. All is going well in using the technique of incorporating the three
functions of the IBS (canonical, time, and dispersion) in a second-level
ANOVA. However, the resulting F-test map, which captures most of the
stimulus event variance reflects both positive and negative events. I
realize that I can set up individual (1 0 0; 0 1 0; 0 0 1 / -1 0 0; 0 -1 0;
0 0 -1) t-test contrasts from the ANOVA, but each of these only captures
variance associated with one of the IBS functions.

Is there a way to construct a directional masks for the canonical function
(i.e., positive or negative activity), while simultaneously including the
non-directional components for the time and dispersion derivatives? Is this
a matter of constructing individual (1 0 0; -1 0 0) masks for the canonical,
and then somehow combining masks for the derivatives with the canonicals?
Something like (1 0 0) + (0 1 0) + (0 -1 0) + (0 0 1) + (0 0 -1) and then
applied to the omnibus F?

Always with the masking questions...

Regards to all,
Jeff

__________ NOD32 2079 (20070224) Information __________

This message was checked by NOD32 antivirus system.
http://www.eset.com