Who Actually Votes in a North Carolina Midterm

What this means for a 2026 campaign#

North Carolina's 2026 general election is a midterm. There is no presidential race on the ballot, so the people who vote will not look like the people who voted in 2024. That affects sheriff, district attorney, clerk, register of deeds, county commission, school board, NC House, and NC Senate campaigns.

The practical rule is simple: build your campaign around voters who turn out in comparable midterm elections. Do not use the registration total as your vote goal. Do not assume a presidential-year turnout rate will repeat in 2026.

We compared the 2022 midterm with the 2024 presidential general for registrants who were eligible for both elections and remain active today. Of 4,818,361 people in that group who voted in 2024, 1,451,173 did not vote in 2022. That is 30.1% of the 2024 presidential electorate.

The people who stayed home were not evenly distributed. Younger voters and unaffiliated voters dropped off more often than older voters and major-party registrants. County turnout also varied sharply. A campaign that copies a statewide rate can miss its own county by a wide margin.

This briefing explains the findings a local campaign can use. The replication package contains the queries behind the figures, and the references page records the outside sources used here. This is our own analysis. It has not been peer reviewed.

The key findings#

1. A midterm has a smaller electorate. Among voters eligible for both elections, 30.1% of people who voted in the 2024 presidential general did not vote in the 2022 midterm.
2. Age is strongly associated with drop-off. Of 2024 voters, 55.6% of people now aged 25 to 34 and 55.2% of people now aged 18 to 24 skipped the 2022 midterm. The rate was 16.8% for voters aged 65 to 74 and 12.5% for voters 75 and older.
3. Midterm voters are older than the registration file. In 2022, 41.3% of voters were 65 or older and 10.3% were under 35. The active registration file is 27.9% age 65 or older and 25.7% under 35.
4. Unaffiliated registrants are the largest registration group, but they turn out less reliably in a midterm. They make up 39.24% of active registration. Of unaffiliated people who voted in 2024, 33.2% skipped 2022, compared with 29.6% of Democrats and 27.1% of Republicans.
5. County turnout has a large range. The measured 2022 rate ranged from 44.9% in Robeson County to 72.9% in Alleghany County. The median was 59.8%.
6. Contested countywide races had very little ballot roll-off. In 2022, contested sheriff contests drew 98.7% of county ballots and contested clerk contests drew 98.4%. If a candidate has a contested race, the larger question is who shows up, not whether voters skip the office once they arrive.

What the data covers#

The analysis uses 9,203,238 North Carolina registration records and 35,525,793 vote-history rows published by the North Carolina State Board of Elections. We loaded the files on 8 August 2026. The State Board publishes a registration file, vote history, registration snapshots, and election results. It does not review or endorse this briefing.

Vote history tells us whether a person participated in a particular election. It does not show who that person supported, how they voted on a ballot question, or why they chose to vote or stay home. Registration party is not a vote choice. Race and age fields are useful for describing the file, but they do not explain a person's reason for voting.

For the main comparison, we used registrants who were eligible on both Election Day 2022 and Election Day 2024 and who are still active in the current registration file. That avoids treating someone who had not registered yet in 2022 as a midterm nonvoter. It also means the results describe today's active voters looking back at past participation. They do not describe every person who was registered at the time.

We use point-in-time registration snapshots for turnout denominators when they are available. That matters because a current registration count includes people who registered after the election being measured. It would make a past turnout rate look lower than it was.

Who is more likely to vote in a midterm?#

Age is the clearest pattern in this analysis. More than half of younger 2024 voters skipped the earlier midterm, while older voters had much lower drop-off. That does not mean every older voter will vote in 2026 or that younger voters cannot be reached. It means a campaign should check turnout history before assuming that its registered voter base will appear at the same rate.

The age mix changes the electorate. People 65 and older accounted for 41.3% of 2022 voters but 27.9% of current active registration. Voters under 35 accounted for 10.3% of the midterm electorate but 25.7% of the registration file. A campaign whose message or volunteer plan depends on a younger registered base should plan for extra voter contact and registration work rather than treating the registration count as support.

Party registration also matters, but it explains less than age in this comparison. Unaffiliated voters are now the largest active registration group: 2,649,690 people, or 39.24% of active registrants. Republicans make up 30.93%, and Democrats make up 29.24%. Unaffiliated voters had the highest midterm drop-off among the three groups.

That is useful for planning, not for stereotyping voters. A campaign should use its own district data, past turnout, and direct contact to decide who belongs in a voter program. Statewide averages are a starting point, not a substitute for a local list.

Turnout changes by county#

North Carolina does not have one midterm turnout rate. The 2022 county range was 28 percentage points, from 44.9% in Robeson County to 72.9% in Alleghany County. The median county rate was 59.8%.

County turnout varies because local conditions vary. Registration age, past participation, population growth, local races, campaign activity, and the top of the ticket can all change the result. This analysis does not identify which factor caused the difference in any county.

For a countywide campaign, start with the county's own comparable election. For a district that crosses county lines, estimate each county's portion separately and add the results. For a local district, use the closest available turnout record and label it as an estimate when the exact district does not have a comparable history.

The main planning mistake is using registration as if it were turnout. Registration tells you the largest possible electorate. A vote goal needs an estimate of the people likely to cast a ballot. The gap between the two is large in a midterm.

Ballot roll-off is usually not the main problem#

Candidates for lower-profile offices often worry that voters will skip their contest. That can happen, especially in uncontested races. The 2022 data suggests a different pattern when the countywide race is contested.

Contested sheriff contests received 98.7% of ballots cast in their counties. Contested clerk of superior court contests received 98.4%. Uncontested races in the same offices received about 71.8% to 76.8% of county ballots.

For a contested campaign, this means the first job is to identify likely voters and make sure supporters turn out. It does not mean ballot education is unimportant. A clear ballot line, useful reminders, and a full-ballot message can still help. The evidence only says that lack of turnout is likely to be a larger problem than broad ballot roll-off in a contested countywide race.

Registration before the deadline#

The voter-registration deadline can affect a close race. In the 25 days before the 2024 general election, 169,815 people who are still active today registered after the statutory deadline. Of those people, 119,058 voted.

Those voters could not have been reached through a standard pre-election voter-registration deadline campaign for that election. A local campaign should know its state's deadlines early, make registration part of its calendar, and avoid waiting until the last weeks to find unregistered supporters.

This does not show how many voters would have registered because of a particular campaign. It shows that late registration is common enough to matter when planning a program.

How to use this in a campaign plan#

Start with a specific contest and a comparable election. A 2026 general-election campaign should usually begin with 2022 midterm turnout, then adjust only when there is a clear local reason to do so. A primary needs primary turnout history, not general-election history.

Next, define the electorate correctly. A countywide race needs countywide turnout. A multi-county district needs every county that contributes voters. A multi-seat race needs the number of seats and past results, not a simple majority assumption.

Then make a voter-contact plan. Use past turnout to build a starting list, talk to voters, and update the plan as you learn more. This data cannot tell you who supports a candidate. It can help you avoid spending a campaign budget as if everyone on the registration file will vote.

Finally, keep the limits in view. This analysis is descriptive. It does not forecast the 2026 outcome, prove why any group dropped off, or predict an individual voter. It cannot see unregistered adults or vote choice. Use it with local knowledge, direct voter contact, and current information from the State Board.

Reading a turnout estimate#

A turnout estimate is a planning tool, not a promise. It gives a campaign a starting point for a vote goal, a voter-contact goal, and a budget. It should be revised when the campaign learns something important about its own race.

Start by asking which election is comparable. A 2026 general-election race is closer to the 2022 general than the 2024 presidential general. A primary campaign should use past primaries. A special election may need a wider range because it can have unusual turnout. If district lines changed, say so and avoid treating an old district result as exact.

Then check what is inside the number. A countywide estimate should include all voters in the county. A district that crosses county lines should include each county's share. A multi-seat contest needs the number of seats and the past result for that kind of ballot. A total from the wrong geography or race type may look precise while answering a different question.

Use the estimate to make a work plan, not to label people as supporters. The voter file can show past participation, but it cannot show who supports a candidate. A voter who usually turns out may not support the campaign, and someone with little history may become an important voter after direct contact. Keep turnout estimates separate from the campaign's own support assumptions.

As the election gets closer, compare the plan with what is happening. Check registration changes, early-voting data when it is available, filed candidates, and reports from volunteers. If the information changes the expected electorate, update the goal and explain why. A campaign is better served by an honest revised estimate than by keeping an old number because it was the first one on the plan.

Methods and limits#

The figures come from the State Board's public registration and vote-history files, registration snapshots for 8 November 2022 and 5 November 2024, and 2022 ENRS precinct results. The queries show the definitions, joins, exclusions, and calculations used for each result.

Some data limits remain. The file has no vote choice and no reason for participation. It cannot separate age effects from generational effects. It does not establish whether a campaign, candidate, policy, or local event caused turnout. Redistricting limits comparisons at the legislative-district level. Results coverage for some local contests is incomplete, so we do not use those records to make statewide claims.

The registration and vote-history files also have ordinary administrative limits. A person who moved, changed status, or was removed and later re-added may not appear in the same way across every snapshot. We keep eligible, currently active registrants for the core comparison and state that rule because it affects the counts.

The work has been checked internally against the published files and the linked queries. It has not been peer reviewed. Treat it as a transparent planning resource, not as a forecast or an independent audit of the State Board.

Sources and replication#

The State Board is the source for the registration, vote-history, and election-result files. The replication package publishes the SQL and supporting scripts behind the figures. The reference list records the source pages and retrieval notes for the outside material used in the analysis.

If you use a number from this briefing, keep its context with it: the election, the geographic area, the voter group, and the method used to calculate it. That is the difference between a useful planning estimate and a number copied out of context.