Partisanship, Race, and the Early Vote in North Carolina

(This post is co-authored with Jacob Canter, Reed sophomore junior and EVIC research assistant)

The early ballots are beginning to pile up in North Carolina, and we can finally start to discern some patterns of turnout by partisanship and by race.

As Michael McDonald has pointed out on Friday, it’s important in this state (as in many states) to carefully discriminate between ballots requested and returned by mail–no excuse absentee ballots–and ballots that are cast in person at an early voting location.   What makes the NC situation a bit confusing for any new to early voting is that the state describes both modes of balloting as “absentee” voting; the early in-person variant is called “one-stop absentee voting” because the absentee ballot is requested “in-person.”  Turnout reports are merged into a single file; the critical field in this file is “Ballot Request Type” which contains entries for “Mail”, “In Person”, “Overseas”, and “Military.”

The first graphic reports the number of ballots returned (and in the case of by mail ballots, accepted) as a proportion of all registered voters, by party.  The interesting pattern to notice here is  the sudden increase in Democratic turnout on the 20th-18th days before the election.  This corresponds to the start of early in-person voting.

The underlying data show that Republicans use by-mail voting at a much higher rate in the state than do Democrats (the first graphic of raw returns is useful in NC only because the proportions of party registrants is roughly comparable–I need to have Jacob reproduce these as proportions of party registrants later this week).


The second graphic shows that of the ballots requested there is only a slight difference in the rate of return by party.  This is what McDonald refers to as the lack of substantially higher Republican enthusiasm in the state at this point.

Finally, the in-person rates by race illustrate both the stronger preference for this method among African Americans in the state, and of course, help account for a large portion of the party pattern observed above.  In my opinion, it’s a bit early to conclude much about Democratic mobilization efforts until we have another few days of data.  We’ll work on comparing these trend lines to 2008 in a few days.