Dear DCMers,
Now
I am engaged with a DCM study using family level inference and BMA. I have two
families: F1 & F2. The exceedance probability of F1 is 0.88. Can I
conclude that F1 is preferred than F2 with high confidence?
Another
question is about BMA. Using BMA, we can obtain model parameters as result. According
to Penny 2010, there are two method to test the significance of these parameters.
One is similar to group analysis in SPM which uses within-subject means as
summary statistic in a RFX analysis. Another is described like this, ‘The
i = 1 … N samples then provide a sample-based representation of the
posterior density p(theta|Y)’ from which the usual posterior means and
exceedance probabilities can be derived’ (Comparing Model Families,
P7, after formula 33). I wonder how to get the exceedance probabilities of the
parameters?
Best
Wishes!
Feng
Lu