Dear SPM list,
Is anyone using the Bayesian option in SPM5?
Below is a post I made a few days ago. I would love to get a response -
even if it's to tell me that my question is inane :-).
Date: Tue, 10 Apr 2007 17:55:16 +0100
Reply-To: Emily Trittschuh <[log in to unmask]>
Sender: "SPM (Statistical Parametric Mapping)" <[log in to unmask]>
From: Emily Trittschuh <[log in to unmask]>
Subject: Bayesian Analysis in SPM5
Content-Type: text/plain; charset="windows-1252"
I have a question about SPM5 Bayesian analysis at the 2nd level. We are
finding that there seems to be a significant difference in the results
based on when contrasts are specified.
My understanding is that contrasts can be entered at the time of the
initial "classical" analysis, i.e., BEFORE Bayesian estimation, or they
can be entered AFTER the Bayesian estimation. ( I think this is different
from 1st level analyses in which, I think, all contrasts are supposed to be
entered before Bayesian estimation.)
We find that even when the same contrast vector is entered, contrasts
that are specified BEFORE the Bayesian estimation appear to show more
voxels as significant (i.e., they pass the gamma threshold) than if the
same contrast is specified AFTER the Bayesian estimation is run.
Have others seen this? Any help in understanding this behavior (is this a
feature or a bug?) would be greatly appreciated.