Doesn't matter. All that matters is that the errors are correlated. If voter turnout is biased in one state, then it likely is in another. The reluctance was just an example.