Source: North Carolina State Board of Elections (NCSBE) absentee-ballot file (one-stop early voting and
mail), the statewide voter-history file, the statewide voter file, and NCSBE voter_stats
registration counts for past cycles. Counts are ballots accepted by the as-of date.
Where 2026 stands
Cumulative curves by early-voting day
Each line is the running total of accepted ballots through that EV day. Day 1 is the first one-stop Thursday; Day 20 is Election Day. Past cycles end at their final count; 2026 stops at the as-of day.
Show data tables for these charts
Cohort retention: who is coming back
Voters are matched across elections by their state voter ID (never published). Every 2026 figure is shown next to the same figure for the previous pair of elections spanning the same gap, at the same EV day.
Show data table
Who is voting: vote history and party switching
The 2026 electorate to date, split by what each voter did in the last comparable election, set beside the 2022 electorate on the same EV day.
Party switching
Counties and congressional districts
2026 against 2022 on the same EV day, all ballots. Party share is of all ballots in the area. Cells with fewer than 10 ballots are withheld and show “<10”, never zero.
Methodology and caveats
Alignment
EV day N is the number of days since the first one-stop (in-person early voting) Thursday, counted from 1. In every cycle Day 1 is Election Day minus 19 days, so each day falls on the same weekday: Day 1 Thursday, Day 17 the final Saturday, Days 18 and 19 Sunday and Monday (no one-stop voting, only mail returns), Day 20 Election Day. In 2018 one-stop voting opened on a Wednesday, a day earlier than the aligned Day 1; those Wednesday ballots are folded into Day 1 so the weekday structure matches. All curves are cumulative, so Day 1 also includes any mail ballots returned before the first one-stop day.
Past cycles are rebuilt from final files
2018, 2022 and 2024 are rebuilt from each election’s final absentee file by the ballot return date, not from daily snapshots. For in-person voting this is near exact. For mail it is approximate: ballots that were later rejected, cured or accepted after a first attempt are placed by their final return date, so a past mail count at a given day can differ slightly from what was reported on that day. Ballots with an unusable return date (one in 2024) are dropped. The 2026 numbers are real daily snapshots of the live file.
One vote per voter
Only accepted (counted) ballots are included, one per voter. A voter who requested a mail ballot and then voted in person counts once, as the accepted ballot. Provisional and Election Day ballots do not appear in the absentee file.
Party and demographics
Party is the voter’s registration party on the ballot record: Democratic, Republican, Unaffiliated, and “Other” (Libertarian, Green and any other). In the cohort tables, “party then” is the party the voter’s ballot carried in the base election, so switchers are visible. Race and ethnicity are as self-reported on the voter record; “Undesignated” is a real category. 2018 absentee records carry no ethnicity field, so 2018 is absent from ethnicity charts. Race and age splits for cohorts use the current voter file.
Turnout-rate denominators
For 2026 the denominator is the live voter file’s active and inactive registrants (NCSBE has not yet published a 2026 voter_stats file). For 2018, 2022 and 2024 it is NCSBE voter_stats registration as published for that election. The two sources are close but not identical, so small differences in rate are not meaningful. Turnout rate is accepted ballots to date divided by registered voters, not a final turnout figure.
Congressional districts
The district table assigns every cycle’s voters to their current (2026) district using the current voter file, so past years are not what each district looked like when it was drawn for that year. About 85 to 97 percent of past ballots match to a current district (lowest for 2018). Registered-voter denominators exist by district for 2026 only, so 2022 district turnout rates show “n/a”.
Cohort retention
The cohort is everyone who cast a ballot in the base election, per the statewide voter-history file. Retention is the share of that cohort with an accepted 2026 ballot by the same EV day. The raw cohort includes people who have since died, moved or been removed; the “still on the rolls” figure restricts to cohort members now active or inactive on the voter file (available for 2026 only). Because both pairs span the same gap in years, the raw comparison is like for like.
Comparability
- 2024 and Hurricane Helene. Western North Carolina counties were disrupted by Helene in 2024 (polling and mail logistics, with emergency rule changes). 2024 is a presidential year with far higher turnout; use it as context, not as the main comparator.
- 2023 mail-deadline law change. A 2023 law ended the three-day grace period for mail ballots arriving after Election Day. 2018 and 2022 curves may include late-arriving mail within that window (placed at Election Day); 2026 will not. This mostly affects the final days of the mail curve.
- Unaffiliated growth. Unaffiliated registration has grown, so shares of the electorate can shift without any change in behavior. Compare turnout rates, not just shares.
Small cells
For counties and districts, and for every cohort cell, a count below 10 is withheld and shown as “<10”. A withheld cell is not zero. This is primary suppression only: a withheld value can still be bounded by subtracting the published cells from a total. No names, addresses or voter IDs are published.