Dear Thilo,
I discussed this with my colleagues - the comparison is valid - the inputs are independent variables and so can be changed and compared.
Kind regards,
Peter
Peter Zeidman, PhD
Methods Group
Wellcome Trust Centre for Neuroimaging
12 Queen Square
London WC1N 3BG
[log in to unmask]
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Thilo Kellermann
Sent: 14 October 2014 09:32
To: [log in to unmask]
Subject: [SPM] BMS for models with different input structure?
Dear SPM-Users,
is is valid to compare different models (DCMs for fMRI data) via Bayesian model selection (BMS) where the models differ with respect to their input structure? More specifically, I modelled three conditions in the "usual" way, i.e. one independent regressor for each condition (A,B and C). Then I modelled these conditions differently, where the first regressor was the sum of all three (A+B+C), the second the sum of two
(B+C) and the last modelled the last condition only (C).
When I perform BMS on several different models for both kinds of input structure, this results in strong support (FFX) for a single model with the last input structure ((A+B+C), (B+C), C). Now I wonder if I can use this result in order to show that this kind of modelling direct and modulatory inputs is superior compared to the independent modelling. So the basic question is if this approach is valid and makes sense, or is it as useless as comparing models with different data, for example?
Thanks a lot for any ideas/comments.
Best regards,
Thilo
|