Variables with no variation during sampling

Questions version agnostic
2

If you check the posterior, you should see that the zi parameter is mostly either 0 or 1:

In [24]: np.mean(trace1['zi'],axis=0)
Out[24]: 
array([0.     , 0.     , 0.     , 0.     , 0.     , 0.9932 , 0.99335,
       1.     , 0.99995, 1.     , 1.     , 1.     , 1.     , 1.     ,
       1.     ])

So you’re not seeing much switching behavior because there isn’t much switching behavior (i.e., there is quite a bit of certainty about the value of this parameter in the posterior). Is that not what you expected?