After three weeks of waiting, we thought every U.S. House race in the country was finally decided. No such luck. FiveThirtyEight has uncalled the North Carolina 9th District1 — an extremely close race that we thought the Republican had won — in light of allegations of voter fraud that, it turns out, may go back to 2016.
On Tuesday, the North Carolina State Board of Elections met with the expectation that it would certify Republican pastor Mark Harris’s apparent 905-vote win over Democratic veteran Dan McCready. But it did not, citing the board’s responsibility “to assure that an election is determined without taint of fraud or corruption and without irregularities that may have changed the result of an election.” Although the state Republican Party objected, all nine members of the board of elections voted to delay the certification. On Friday, there was no sign that the race would be certified anytime soon; the board of elections voted to hold an evidentiary hearing on or before Dec. 21.
The week of the election, investigators began looking into possible absentee-voter fraud in Bladen and Robeson counties, in the 9th District’s less populated eastern end. On Thursday, local media reported on several affidavits submitted by the North Carolina Democratic Party in which voters said that people came to their doors to collect their unsealed absentee ballots. One voter alleged that her ballot was incomplete at the time and that the collector said “she would finish it herself.” Another claimed to have received an absentee ballot that she did not request.
FiveThirtyEight also took a look at absentee-by-mail ballots in the 9th District and found that the absentee statistics and the election results do imply that something unusual happened in absentee-by-mail voting in Bladen County (and, to a lesser extent, Robeson).
In six of the district’s eight counties, fewer than 3 percent of total votes were absentee-by-mail ballots. But in Bladen County, that number was 7.3 percent, suggesting a push to get people there to vote by mail. Absentee-by-mail ballots tended to be much better for McCready than other ballots — except in Bladen County. There, absentee-by-mail ballots were more Republican-leaning than the rest of Bladen’s votes by 8 percentage points. But if we look at the whole district, absentee-by-mail ballots were 24 points more Democratic-leaning.
Something looks off in Bladen County’s absentee ballots
Share of votes that were absentee-by-mail and margins by county as of Nov. 30, 2018
| Ballot Margin | ||||
|---|---|---|---|---|
| County | Absentee-by-mail Share | Absentee- by-mail |
All Other | Difference |
| Bladen | 7.3% | R+24 | R+16 | R+8 |
| Robeson | 2.2 | D+21 | D+15 | D+6 |
| Scotland | 2.3 | D+21 | D+13 | D+7 |
| Cumberland | 2.2 | D+16 | D+4 | D+12 |
| Mecklenburg | 5.8 | D+29 | D+8 | D+21 |
| Richmond | 2.0 | D+29 | D+2 | D+27 |
| Anson | 2.0 | D+49 | D+16 | D+33 |
| Union | 2.8 | D+21 | R+21 | D+43 |
| Districtwide | 3.8 | D+23 | R+1 | D+24 |
Particularly troubling is the fact that only 19 percent of Bladen County’s accepted absentee-by-mail ballots were cast by registered Republicans; 42 percent were cast by registered Democrats, and 39 percent by unaffiliated voters.
Michael Bitzer, a political science professor at Catawba College who studies North Carolina politics, told the Charlotte Observer that it’s hard to square that statistic with the fact that Harris received 61 percent of Bladen’s absentee-by-mail vote. As Bitzer noted in his analysis, Harris would have had to win every registered Republican, along with nearly every unaffiliated voter and some registered Democrats, to reach that total, which seems implausible. In other counties, McCready’s margin in absentee-by-mail ballots tracked closely with the percentage of absentee-by-mail ballots cast by registered Democrats.
Bitzer also found that Robeson and Bladen counties had unusually high percentages of requested absentee-by-mail ballots that were then never returned — 40 percent in Bladen and 62 percent in Robeson, while no other county exceeded 27 percent. By raw numbers, Robeson’s 1,180 unreturned ballots far exceeded even the 762 in far more populous Mecklenburg County, which cast more than three times as many total ballots as Robeson.
It turns out that the data irregularities in Bladen County extend to past elections, too. The 9th District saw a contentious and razor-close Republican primary in May 2018, when Harris ousted incumbent Rep. Robert Pittenger by 828 votes. Bladen accounted for 56 percent of the absentee-by-mail ballots (456 of 811) cast in that primary despite being home to just 6 percent of those who voted — and Harris won 437 of those absentee-by-mail votes to Pittenger’s 17, a massive gap nowhere near either the absentee-by-mail results in other counties or the other results in Bladen. And in the 9th District’s 2016 Republican primary, 22 percent of the race’s absentee-by-mail ballots were cast in Bladen County. That time, they went disproportionately to Pittenger challenger Todd Johnson — he won a whopping 221 of Bladen’s 226 absentee-by-mail votes.
None of this necessarily means that there was wrongdoing in the North Carolina 9th. True voter fraud, remember, is quite rare. High levels of absentee-by-mail ballot requests and high candidate performance among those ballots could also be a function of strong get-out-the-vote programs run by the campaigns of Harris and Johnson. Still, given the irregularities in absentee-by-mail votes in Robeson and Bladen counties, we should take seriously the possibility that voter fraud occurred here.
If it did, the big question is whether fraud actually swung this incredibly close election. In Bladen County, where the evidence suggests that too many absentee-by-mail ballots were cast, only 684 absentee-by-mail ballots were accepted; they went for Harris 420-258.2 If fraudsters did indeed collect absentee-by-mail ballots in Bladen County in order to mark them off for Harris, it didn’t produce enough illegal votes to cancel out his 905-vote margin districtwide. But in Robeson County, 1,180 absentee-by-mail ballots that were requested were never returned. If some of those ballots were destroyed, that is a big enough number that McCready could theoretically have gained 906 votes — although it’s still unlikely. Plenty of absentee-by-mail ballots never get returned for one reason or another (lost in the mail, people get too busy, etc.), even in a normal election. But with 1,364 absentee-by-mail ballots cast between the two counties (679 for Harris), and 1,675 absentee-by-mail ballots requested and not returned, it may be impossible to conclude whether voter fraud across both counties changed the election outcome.
Where do we go from here? It’s unclear. If the state board of elections finds that fraud did indeed taint the results, it has the authority to call a new election — even if the fraud wasn’t widespread enough to change the winner. And according to the U.S. Constitution, the soon-to-be-Democratic House of Representatives has the final say on the qualifications of its members; it could theoretically refuse to seat Harris until it’s satisfied that he won fairly.