His argument was that juries prevent "arbitrary court system to throw people in jail who are innocent and removes other protections like jury nullification". There is nothing observable that would confirm that.

They don't do that, they prevent pretty much nothing.

It says nothing one way or the other because the system isn't being used in those cases.

Put simply, if you want to make claims about whether jury trials or bench trials put more people in jail wrongfully by citing US incarceration data you have to somehow contend with the fact that the vast majority of cases don't go to a trial of any kind. Or at least have an argument as to why switching to bench trials would change that circumstance.