For each county listed below is a page with per-precinct results. These pages were generated using the raw results [ZIP, 1580 KB] and the raw voter history/turnout statistics [ZIP, 4148 KB] from the North Carolina State Board of Elections website.
IMPORTANT: The data from the State Board tends to be inconsistent and generally wrong. There is not necessarily any consistency -- even in naming the precincts -- between the two data files. This creates a situation where a voter (and therefore their turnout) is listed as being a "curbside" ("C") or "transfer" ("T") vote, but the actual vote they cast is recorded as coming from the (presumably) correct precinct.
The State Board gets their data from the counties, and seem either unwilling or unable to force the counties to make sure the data -- as reported to the State Board -- actually aligns properly. Presumably the counties have the correct data locally, but I did not have the time or willingness to badger each county individually for correct data. This is what one election official told me about the process:
"The SBOE gets their information from the counties. One reason the data may keep changing is when a voter is removed from a county, their voter history is also removed. If a voter moves to a new precinct within county, the voter history also moves, so precinct 'old' will lose a vote and precinct 'new' will gain a vote.
"We voted X ballots and only show Y voter histories. The reason for the shortage of Z voter histories is not fraud. [Numbers hidden to keep election official anonymous. --jdm] There are several possible reasons--keep in mind that we input voter history in late Nov and Dec, well after Canvass. We scan the labels on the ATVs from each precinct into our system to create the persons voter history. This is time consuming and labor intensive. Occasionally, two pages stick together or a section is missed. Also, this year we had some precincts put the labels on the curbside application and not the ATV so we missed inputting them. Lastly, the scanner, when it properly reads the label, makes a beeping sound--sometimes we go too quickly and one voter is not complete and we've overridden them with the next one. Bottom line, it is human error. I made the management decision not to go back and look at every single page and try to make this number perfect. The number of ballots cast and counted is a much more critical number."
NOTE: There are many places where the turnout for a precinct is listed as "1"; this is my own notation to indicate that there was no turnout available for that particular precinct.
IN SUMMARY: Occam's Razor says that it is more likely that the local data is correct and the state has an incorrect copy than it is that fraud and overvotes took place in one-quarter of the precincts across the state. Contact your local county BoE to verify turnout -- and the corresponding undervote/overvote rates -- before publishing a breathless article about "widespread fraud" in North Carolina.
Thank you.
-jdm