Hmm, tricky question - too many unknowns.
But let us assume that D makes up 35% of the voters, then 45% x 35% is about 16% which is 50% of the early voters.
If we then assume that I makes up 30% of the votes, and only a third of them has voted early and furthermore that they vote 55 - 45 for Romney, then they will make up 30% x 45%x 33% of the D vote in the early sample = 14%
D: 50% + 14 = 64% which is pretty close to what RAS got in the sample of the early voters.
Of course looking at the whole picture with a GOP 35% and the I’s splitting 55-45 in favour of R it looks good!
every person I know who has voted early has voted for Romney,so I don;t know about mostly the left votes early