Kruschke’s BEST model for multi-group comparisons

[romainmartinez]

Ok makes sense. So i could just substract the trace like so:

pm.plot_posterior(trace['method[T.y]'] - trace['method[T.z]'], ref_val=0)