Dear Alex,
See responses below.
All the best,
Will.
________________________________________
From: Alexander Taylor <[log in to unmask]>
Sent: 03 October 2015 11:03
To: [log in to unmask]; Penny, William
Cc: Alexander Taylor
Subject: Re: Bayesian model estimation for multiple sessions
Hello Stephen/Will,
Thanks for the responses and further updates to my questions.
So if I am understanding the conversation correctly, in order to correctly run first level bayesian estimation for runs or sessions I do not select the matlabbatch{2}.spm.stats.fmri_est.method.Bayesian.anova.first = 'Yes';
>> That's correct. You'll then be able to have multiple sessions.
instead I choose second level correct?
>> You can choose ANOVA->second level->'Yes' in the GUI. SPM will then automatically write out contrast images for the main effects and interactions implied by the
specification of your factorial design. But you don't need to say Yes here, as you can always create the con images yourself. You will then take these con images (whether created automatically
or not) and do a second level Bayesian analysis if you wish.
Or is it that this is just simply running a second level analysis on already computed lower level first level contrasts from classical modelling and computing second level bayesian estimates for comparisons between different sessions and not between runs?
Many thanks
Alex :)
|