Hello,
I performed a DCM study in patients and a control group. While BMS for patients found 1 winning model (RFX), BMS on controls (no families, individual models) selected two models (2,9) amongst 9 possible models to be superior in explaining the data. No I wanted to caclulate averaged parameter estimates over these 2 models using BMA.
Could you please check whether these next steps would be correct?
1. Redo RFX BMS with 2 Families: 1: models 2,9. 2:1,3:8. Set BMA to be computed for family 1.
2. Use BMS.DCM.rfx.bma.mEps (subjects.B) for the individual subjects for further statistical inference (correlations etc.).
Thank you very much for your help.
Best, Robert
|