I wonder how much real difference you get in the statistical sampling margin of error when going from 58% to 70%. How much accuracy does that buy in the election outcome for the extra time that additional 12% of the population spent?