I'll admit going in that I haven't read the article, but I have to question this just on logic. How was this poll taken? What guarantees are there that the respondents were actually early voters? Even if that's true, how were they determined to be representative of the whole of early voters? Was this just a snapshot of early voters on one or two days, or were they actual exit polls all during the early voting period? Frankly, I have serious doubts that any polling procedure could get an accurate read on early voting results.