I found this phenomenon while trying to simulate common mode interference across an optocoupler. Turns out this was difficult, since there was common mode coupling even without any circuit element connecting the two parts. Even without the optocoupler! I have narrowed it down to this minimal example. My guess is that the problem lies in the opamp model. I have tried various models, from the most simple to the proprietary LT6268. There are variations in the coupling between models but none of them seems to work as expected. Anyone seen this before? As you can see in the circuit, the ground node (node 0) has one single connection. It sets the reference voltage for the rest of the circuit but with its single connection it should not be able to affect the circuit in any other way.