How to do correlation the bayesian way with two different sized elements?

Yes, I think you’re right. I was “grasping at straws” hoping there was a way to avoid that.