Hi SPM users, I tried to use spm_dcm_sessions to run dcm random effect analysis of 8 models for comparing 2 connections in the B matrix. The routine ran successfully without any error message. In Mablab window, the mean differences of the connections and the variances of the models were displayed. I found that all the means and variances were equal to the mean and variance of the 1st selected model. I checked spm_dcm_sessions, I think there may be a mistake in line 53: "load(deblank(P(1,:)))". Should it be "load(deblank(P(model,:)))"? Secondly, it is valid to use "contrast of connections" to compare the connections in an averaged model? If yes, what are the differences between using this method and using the fixed effect analysis of individual models implemented in spm_dcm_sessions? Thanks in advance for your answer. HoMing.