Virginia's Redistricting Referendum: A Partisan Map with a Demographic Twist

What Happened on April 21

Virginia voters approved the Democratic-sponsored constitutional amendment

allowing the General Assembly to redraw congressional districts for the

remainder of the decade. The margin was 1,575,288 Yes to 1,486,229 No —

51.45% to 48.55%, a 2.9-point Yes win with 133 of 133 localities reporting.

That narrow margin is the story. Democrats got their map. But the Yes

coalition was significantly smaller than the one that elected Governor

Abigail Spanberger five months earlier, and the shape of where support

came from diverged from both the 2025 gubernatorial and the 2024

presidential results in ways worth understanding heading into the 2026

House midterms.

At the precinct level, the pattern is dense: 2,514 of 2,534 Virginia

precincts, coloring each by its Election Day Yes–No margin.

!Virginia referendum Yes-No margin by precinct — Election Day votes

Blue precincts (Yes > No) cluster in exactly the three places Virginia

Democrats always cluster: Northern Virginia's I-66 and I-395 corridors,

the Hampton Roads / James River core, and Richmond-Charlottesville. Red

dominates everywhere else. Yes won only 38 of 133 localities but

carried the state because those 38 include almost all of its population.

The partisan map barely moved. At the locality level, the Yes–No margin

correlates with Spanberger's D–R margin at r = 0.998 and with Harris's

D–R margin at r = 0.996. Every locality that voted D for Spanberger

voted Yes. Every locality that voted R for Earle-Sears voted No. Only a

handful of localities switched sides, all tight.

What changed was who showed up and how hard they pulled. That's where

the coalition story lives.

A 12-Point Gap the Statewide Result Hid

If you zoom out to statewide margins, the three elections look like a

single trend — three snapshots of an increasingly Democratic Virginia:

But line them up against each other at the locality level and two

asymmetries jump out:

Against Spanberger, Yes lost ground everywhere — no single locality

outperformed her margin. Against Harris, one region stands out: **Yes ran

ahead of Harris across Northern Virginia, and ran behind Harris in

every other region**. Those two results encode different stories about

what the April 2026 electorate actually was.

!Margin by region: Harris 2024 → Spanberger 2025 → Yes 2026

Every region moved D from Harris to Spanberger — a personal-brand

overperformance. But every region except NoVa moved back toward the

partisan middle for the referendum. NoVa held. Understanding why means

looking at who was in the April voter pool.

Where the Composition Changed: Hispanic NoVa

The I-66 and I-95 corridor outside DC is one of the largest Hispanic

population concentrations in the country. Manassas Park is 47% Hispanic.

Manassas City is 44%. Prince William County is 26%. Harrisonburg — down

the Valley but culturally analogous — is 24%.

In 2024, Harris lost ground with Hispanic voters nationally, and Virginia

was no exception. Spanberger recovered that ground in November 2025 with

a moderate, broad-coalition D-base campaign. But in April 2026, low-propensity

Hispanic voters — the marginal Harris coalition voters Spanberger partially

pulled back — mostly stayed home. The electorate that remained was

smaller, older, whiter, wealthier, and more Democratic-reliable.

Result: Yes ran ahead of Harris in exactly those places where Harris's

2024 performance was weakest.

The locality-level pattern is monotonic. Across the 129 localities with

complete data, Hispanic share of population has the strongest

correlation with Yes's swing against Harris of any demographic variable

we measured (Pearson r = +0.75).

!Hispanic share vs Yes swing vs Harris 2024

The quintile breakdown makes it starker. The most-Hispanic fifth of

Virginia — 9.3% to 46.8% Hispanic, containing roughly 1.4 million

referendum votes — was the only quintile where Yes outran Harris.

Every less-Hispanic quintile ran behind Harris, and by more as the

Hispanic share fell.

The Asian Pocket: Loudoun, Fairfax

Loudoun County is 22% Asian. Fairfax County (including Fairfax City) is

19-20% Asian — the densest Asian American population in the

Mid-Atlantic. These are also the counties where Harris's 2024

performance relative to Biden's 2020 was weakest among Asian voters,

especially among Indian and Vietnamese subgroups.

Same composition story applies. The April 2026 electorate had far fewer

of the low-propensity Asian voters who had held for Biden and drifted

for Harris. What remained was a more Dem-reliable core.

The correlation of Asian share with Yes swing vs Harris is r = +0.61

— second only to Hispanic share. Loudoun, despite only 64% of its

2024 turnout showing up, delivered a 5-point improvement on Harris's

margin.

Rural White Virginia: The Counter-Mobilization

The opposite story plays out in the Valley and Piedmont. Rockingham

County — 84% non-Hispanic white, median age 40, Harrisonburg's

agricultural hinterland — ran **13.5 points behind Harris on the

referendum**. Rockbridge, Augusta, Highland, Bath, Smyth, Wise: all

followed the same pattern.

These localities ran 72–77% of their Pres 2024 turnout — more than

NoVa's 63–65%. The Republican side wasn't depressed; if anything, R

counter-mobilization on an explicit "stop the Democratic gerrymander"

message brought out voters who sit out presidential elections. The No

share in the Valley looks like the R share in a hot midterm, not a

sleepy April special.

Correlations tell the same story. Non-Hispanic white share correlates

with Yes swing vs Harris at r = –0.54. Median age correlates at

r = –0.50. The April 2026 electorate in rural Virginia was whiter,

older, and more anti-amendment than the Trump electorate had been

pro-Trump five months earlier.

The College-Town Exception

There's one compositional pattern that hurts Dems in both comparisons:

student turnout collapses in April specials. Four localities where

Spanberger famously overperformed on student turnout got hammered in

the Yes column:

Montgomery County flipped. Radford City flipped. Harrisonburg held but

shed 15 points of margin. These places move on the student vote, and

the student vote can't be mobilized in April when exams and

commencement are weeks away. This is not a durable realignment — it's

an off-cycle scheduling artifact that Democratic House candidates on

November 3, 2026 should not be modeled from.

Hampton Roads: The Dem Machine Held

A fourth pattern deserves attention: **majority-Black Hampton Roads

and Richmond cities held up best against both baselines**. Petersburg

(77% Black) ran only 3.4 points behind Spanberger — the smallest

underperformance of any locality. Portsmouth (51% Black), Hampton

(49%), Newport News (41%), Norfolk (40%), Suffolk (42%) all ran within

10 points of Spanberger and matched or exceeded Harris.

This is a turnout infrastructure story. Cities with established Black

Democratic political operations — ward clubs, churches, labor

affiliates — kept their base voters engaged even for an April special.

Rural majority-Black Southside tells the opposite story: Charles City

(41% Black) dropped 8.8 points vs Harris on 68% turnout; Sussex (54%

Black) and Halifax (35% Black) both underran by 7-8 points as

low-propensity Black voters stayed home. The Dem machine stops at the

Richmond metro line.

What This Means for the 2026 House Midterms

The partisan map on the Virginia referendum is almost indistinguishable

from the Spanberger map. That's superficially reassuring for Democrats:

it confirms the Spanberger coalition is stable as a partisan map, not

a cult of personality. But the **underperformance vs Spanberger is

large and non-random** — 12 points statewide, concentrated in the

exurban counties (Spotsylvania, Culpeper, Fauquier, Louisa, Orange)

and college towns where Spanberger's margin was softest.

Three implications:

**1. Midterm Dem turnout in VA will look more like the April 2026

electorate than the November 2025 electorate.** The composition that

flatters Yes in NoVa — Dem base voters without the marginal Harris

coalition — is the same composition that shows up in a House midterm.

That's good for the 4 new D-leaning seats the map creates; it's less

good for any candidate who needs to pull Spanberger-style

overperformance in exurbs.

**2. The Hispanic/Asian vote is elastic — and turnout-dependent, not

persuasion-dependent.** The Q5 Hispanic quintile outperformed Harris by

registering fewer weak Harris voters, not by winning new voters. To

replicate Spanberger's full margin in 2026, Dems need those marginal

Hispanic and Asian voters physically in the November electorate.

Organizing, not messaging, is the variable.

3. Rural R turnout in Virginia is not dormant. The Valley, Southside,

and SWVA delivered 95%+ of their Gov turnout for an *April special on an

obscure constitutional question*. In a nationalized House midterm, rural

R turnout should be modeled at full November strength, not dropped off.

For the full interactive map, a locality-by-locality table, and the

coalition-vs-baselines view, see the dedicated race page:

/elections/2026/measures/virginia/redistricting-amendment.

Methodology

Referendum results pulled from the Virginia Department of Elections

ENR system (enr.elections.virginia.gov) at 2026-04-22 18:51 UTC, with

133 of 133 localities reporting. Presidential 2024 and Gubernatorial

2025 results are from the Akashic Edge elections database, with source

provenance to VA Department of Elections official certifications.

Precinct-level caveat. Virginia publishes precinct-level results

for Election Day votes only (2.63M of 3.06M total ballots). Early

voting, mailed absentee, and provisional ballots — roughly 14% of the

statewide vote — are reported at the locality level only. Our precinct

map above shows Election Day votes exclusively; all statewide and

locality-level margins throughout the article include every mode.

2,514 of 2,534 precincts matched to 2024 Census boundaries (99.2%);

20 Bristol/Goochland/Williamsburg precincts didn't match due to

mid-cycle precinct renaming and are excluded from the map.

Four Virginia county/independent-city pairs with matching names

(Fairfax, Franklin, Richmond, Roanoke) were collapsed for the

comparison analysis to match a known ETL merge in the 2025 Gov load —

129 matched localities in the final comparison dataset.

All margins are computed as (D_votes − R_votes) / total_votes × 100

(or (Yes − No) / total for the referendum), following Akashic Edge's

standard margin convention. Demographic data is 2020 PL 94-171 Census

redistricting data plus ACS 2024 5-year estimates.

Correlations are Pearson, weighted at the locality level (not by vote

volume). Quintile rollups are weighted by referendum vote volume.

Complete 129-locality dataset available at

akashicedge.com/historian.