Dear Pari,
The (free-energy approximation to the) log-evidence is stored in DCM.F. When plotting the log-evidences of alternative models in BMS, the values are normalised by subtracting the model with the lowest log-evidence. This is the reason for the difference you observe.
I hope this helps!
Martin
> On 04 Apr 2016, at 18:23, Pari Eghbali <[log in to unmask]> wrote:
>
> Dear SPM experts,
>
> I have performed BMS within each subject. For group report, I want to draw a general graph shows my three models log evidence and accuracy over 8 subjects as like as some papers, so I need log evidence and accuracy of them but F in BMS of each subject is different with log evidence of graph. How can I obtain log evidence of each model?
>
> Thank you for the response,
> Pari
|