Which Voting Rules Could Actually Change the Midterms?

The calendar separates two very different election stories. North Carolina voters face a November 10 deadline to correct certain ballot defects under legislation already enacted this summer. The Supreme Court’s major Arizona citizenship-registration case, meanwhile, is scheduled for argument on December 8, more than a month after the midterms. Both concern access to the ballot. Only one of those dates describes an immediate hurdle for November voters. (ncleg.gov)

That distinction should guide how we assess voting-rule fights as of October 3. I put North Carolina’s registration-repair and ballot-correction procedures, along with Pennsylvania’s implementation of mail-ballot protections, ahead of speculation about a sweeping Supreme Court intervention. They have identifiable voters and operating instructions attached. But the evidence supports targeted scrutiny, not a prediction that restrictions will deliver particular seats to Republicans. (ncsbe.gov⁴)

There is a serious reason to resist those predictions. Voters and campaigns adapt. Economists Enrico Cantoni and Vincent Pons, studying strict identification laws across states, estimated an average turnout effect of minus 0.1 percentage points in their demanding specification. Campaign contact with nonwhite voters increased by 4.7 points, suggesting mobilization might offset burdens. Yet the turnout estimate’s uncertainty interval ran from minus 3.0 to plus 2.8 points. That is evidence against assuming automatic disenfranchisement, but far too imprecise to certify harmlessness in close contests. (academic.oup.com)

North Carolina offers a concrete test. Its Registration Repair Project seeks missing driver’s-license or Social Security identification numbers in existing voter records. The state describes an approximately 82,000-person mailing cohort. That number is neither a current count of unresolved records nor an estimate of people who will lose their votes. The state’s lookup tool updates daily. (ncsbe.gov⁴)

Voters can supply the information online, in person or through prepaid mail. Those whose records remain unresolved must vote provisionally, meaning their ballots are held while officials determine whether the counting requirements have been met. They can provide identification information during that process; alternatives exist for people who have neither identification number. Calling this an 82,000-voter purge would misdescribe the program. The electoral question is how many otherwise-participating voters fail to complete it, and where they live. (ncsbe.gov⁴)

The ballot-correction rules deserve equally close attention. North Carolina’s September 4 instructions prohibit third-party submission portals and invalidate cure documents returned by unauthorized people, while preserving disability assistance and direct submission by email, fax, mail or in person. Ballot curing is the process of correcting an eligible defect so an already-cast ballot can count. A portal ban therefore does not mean electronic curing has disappeared. The relevant loss would come from voters who cannot successfully switch to an authorized channel. (ncsbe.gov⁵)

Timing can make that distinction unforgiving. Cure documents must arrive by noon on the fifth business day after Election Day: November 10 this year. Officials must promptly notify voters of curable defects. For defects that require a replacement ballot, however, discovery from the Saturday before Election Day onward triggers instructions to vote in person if time remains. A legally available second chance can become practically useless when notice arrives too late. These are current operating conditions; the September revision date alone does not establish that every restriction is new. (ncsbe.gov⁵)

Even an enacted law requires provision-by-provision reading. North Carolina’s August legislation establishes the cure deadline and extends the timetable for specified ballot tallies from the third to the fifth business day after the election. But its new affidavit requirement for relatives or guardians delivering absentee ballots does not begin until January 1, 2027. It should not be counted as a 2026 voting burden. (ncleg.gov)

Pennsylvania presents a different immediate priority: whether protections won in court actually reach voters. The state’s September 30 guidance says ballots in undated or incorrectly dated return envelopes must be canvassed, while requiring their segregation. It also tells counties to notify voters when a known defect will prevent counting, allowing correction where offered or provisional voting on Election Day. Those instructions create opportunities to save votes, rather than merely new reasons to reject them. (pa.gov⁶)

The dating dispute is not over, but its procedural status matters. The Supreme Court has been asked to review the federal ruling protecting these ballots. Its dockets show a June 29 request for the federal government’s views, not an agreement to hear the cases. Pennsylvania’s current guidance continues to require counting. Meanwhile, the ordinary civilian mail-ballot receipt deadline remains 8 p.m. on November 3; a timely postmark alone is insufficient. (supremecourt.gov⁹)

Administrative data establish the scale without establishing a partisan outcome. Pennsylvania reported that roughly 1 percent of returned mail ballots were rejected in the 2024 general election. Date errors accounted for 23 percent of rejections, late receipt for 33 percent. These are distinguishable problems requiring different remedies. A victory over the date requirement does nothing for a ballot delivered after the receipt deadline. (pa.gov)

Nor should every national lawsuit remain on the danger list indefinitely. On June 29, the Supreme Court decided Watson v. Republican National Committee, rejecting the claim that federal election-day statutes prohibit Mississippi’s post-election receipt window for timely postmarked ballots. That decision preserves state flexibility; it does not extend Pennsylvania’s deadline. Arizona’s separate citizenship case was accepted that same day, but its December argument schedule places the ordinary merits process beyond November voting. Neither case justifies assuming an imminent nationwide rewrite. (supremecourt.gov¹³)

The case for watching actual ballot failures remains compelling. Brennan Center researchers studying Texas’s new identification-number requirement found that about 30,000 of 215,000 mail-ballot applicants in the 2022 primary had an application or ballot rejected. Nearly 90 percent of that rejected group did not ultimately participate. That descriptive result does not measure North Carolina’s causal effect, and a first-use primary is a poor substitute for a later general election. It does demonstrate that another voting method on paper does not guarantee recovery. The researchers also reported slightly higher ballot-rejection rates for Republicans, undermining any automatic assignment of the losses to Democrats. (brennancenter.org¹⁴)

To translate such failures into election consequences, I would use explicit arithmetic rather than the largest available mailing list. Suppose, purely as a stress test, a rule costs 2,000 counted votes after all repairs and replacement voting. If those voters would have divided 60–40 between two candidates, the margin changes by 400 votes. At 50–50, it changes by zero. A 400-vote movement can reverse a 100-vote result, but not a 1,000- or 10,000-vote result. These are scenarios, not projected margins or measured effects.

Small margins have real precedents: North Carolina’s House District 105 was decided by 213 votes in 2024 and is contested again this year. (axios.com¹⁵) But statewide exposure cannot simply be assigned to that district. Nor should every North Carolina contest be treated as equally close: Inside Elections rates its Senate race Lean Democratic and its 1st Congressional District Tilt Republican. Neither rating supplies a numerical margin forecast. (insideelections.com¹⁶)

I would therefore keep candidate strength and mobilization as the baseline explanation while treating unresolved registration records and failed ballot repairs as concrete, localized risks. The missing evidence is district-linked failure after recovery, compared with similar voters’ participation and defensible estimates of candidate preference. We do not need to pretend those estimates already exist to take the problem seriously. And protecting votes has value even when no winner changes: an ACLU of Pennsylvania analysis found that more than 9,000 voters cast counted ballots in 2024 after first making disqualifying mail-ballot mistakes. (aclupa.org)