In North Carolina, one-stop voting allows people to register to vote and to vote early at the same time. In 2008, 251,893 voters registered during one-stop early voting. These 250 thousand new registrations included 104,355 new voter registrations (xls) and 147,538 updated voter registrations (xls).
In total, they accounted for 5.8% of all voters. For context, about 60.6% of North Carolinians voted early in 2008.
Using official turnout statistics and exit polls, it's possible to estimate what Obama and McCain won from these voters, even though we only have their Party ID (and gender). The methodology is below the fold.
Obama received an estimated 147,588 votes and McCain 102,962, for a pro-Obama margin of approximately 44,626. Considering the more than 4 million total voters, these 45 thousand votes improved Obama's margin by ~1.0% of the overall vote.
These voters would not be captured by pollsters unless interviewed after registering and voting at the same time. Michael McDonald (who organizes all sorts of great election statistics out of George Mason University) recently wrote:
Unregistered voters don't even make registered voter poll screens, much less likely voter screens. It will be worthwhile to watch if one stop voting moves the North Carolina polling as early voting progresses.So, the polls in North Carolina could narrow somewhat in the coming weeks solely as a function of these newly registered voters being included in the polling. But these voters could also be more likely than other people to only have cell phones, and so automated pollsters may not pick-up as much movement as those that call cell phones.