Election day has come and gone, yet media outlets have not projected a winner in several statewide races.
The biggest, obviously is the presidential race. According to the North Carolina State Board of Elections (SBE), as of 2:30 PM on the day after the election, President Donald Trump leads form Vice President Joe Biden in North Carolina by 76,701 votes. Similarly, Senator Thom Tillis leads Cal Cunningham by 96,707 votes.
Media outlets are hesitant to declare a winner in those races because the SBE says that 117,000 absentee-by-mail ballots are still outstanding as of 7:30 PM on election day.
However, Trump and Tillis’ leads are safe.
The SBE noted that the 117,000-outstanding absentee-by-mail ballot projection “does not yet account for voters who cast a ballot on election day.” That also does not account for people who requested absentee ballots but did not submit or mail them by 5:00 PM on election day. In short, the number of mail ballots that will be accepted by election officials by the November 12 deadline will be much less than 117,000.
So, how many more ballots can we expect to be accepted?
We cannot say with precision, but we can make a projection based on the pattern of returns so far this election and a comparison with the post-election day pattern of mail ballot returns in 2016. As with all projection models, there are some assumptions built in that could cause the model to vary from the actual results, so proceed with caution.
Over the last ten days of the election period, including election day, 194,165 mail ballots were accepted this year compared to 75,520 in 2016, a difference of about 2.57 times. Assuming that a similar difference in returns will take place after election day in 2020 compared to 2016 post-election day mail votes, we can project that about 25,625 absentee-by-mail ballots will come be accepted after election day if the rules governing ballot acceptance were the same.
However, they are not the same. Due to a collusive settlement between the SBE and Democratic attorney Marc Elias, the deadline for ballots mailed from the United States has been changed from three days after election day to nine days after election day (November 12). How will that affect the projection?
Digging through one-stop and absentee-by-mail data from 2016, I found that 671 ballots received by election officials more than three days after election day in 2016 were marked as being late. Not all of those ballots would have been accepted with a later acceptance date; some may have been missing a postmark and others may have been postmarked after election day. However, we will assume for this model that all of those late ballots would have been accepted if the deadline to receive them was nine days instead of three. Using the same 2.57 difference in rate for comparing those numbers that we did for projecting the non-late ballots we can project an additional 1,724 ballots been accepted due to the change in the acceptance date for this year (and, for reasons stated above, that is almost certainly too high), we can project that roughly 27,349 ballots will be accepted after election day.
We can even project what days those ballots will come in if the pattern of submissions in 2020 will be similar to 2016 and account for the “late” ballots that will start coming in four days after election day:
Base on this projection, Trump and Tillis’ leads are safe. The only statewide candidates whose leads are smaller than the projected number of mail ballot returns are Democrat Josh Stein, who is 10,769 votes ahead of Jim O’Neill in the race for attorney general, and Republican Paul Newby, who is 3,742 votes ahead of Cheri Beasley for chief justice of the NC Supreme Court.
However, if post-election absentee ballots break for Democrats at the roughly 2-to-1 rate that previous mail ballots have done, the only leading candidate who is in danger of losing is Newby.
So you have what I believe to be a realistic projection of how many ballots we can expect to come in between today and November 12. As with all models, there are assumptions built into this one, but they are reasonable and fit within established patterns of voting behavior.
UPDATE: The actually absentee-by-mail post-election turnout was even lower than the model projected, with under 15,000 ballots coming in after election day.