 |
 |
Dear all, I am a little confused about the result of my fixed effects BMS of 8 models in 8 subjects. When looking at the group result, model 2 appears to win by >100, and has a posterior probability of ~1. When I examine the BMS.DCM.ffx.F matrix, and when I minimum correct the values across my subjects I find that model 2 only wins in 2/8 of my subjects. In one of my subjects model 2 does very very well relative to the others, which makes me think this subject is biasing my group result. What is really odd is that in that subject, when I look at the mEps.B matrix, the connections purportedly modulated in the winning model have b-values of 0??? I have examined the DCM.xY.u values and there doesnt appear to be anything wrong there. Similarly, the range of the A-parameter values for that subject are ~0.15Hz, compared with the mEp A-parameter range of ~0.1Hz. Any suggestions? Thanks Josh