V-Map · U.S. election history · 2006–2026
Election data sources and review methodology
V-Map preserves the distinction between reported vote counts, actual ballots, population references and comparisons that the available sources can support.
Import and source reconciliation
Source manifests identify publishers, URLs, retrieval or version dates, hashes when available, and upstream terms. Explicit source priority resolves overlapping records while preserving source-selection evidence. Known defects, missing counts, incompatible boundaries, incomplete scope and unqualified ballot rounds receive holds rather than invented values. Democratic, Republican and Other labels retain original source-party information; Other votes are never assigned to a major party for a comparison.
Actual reported county results are required. House county totals are constructed only when every published county-district portion has been verified. District-wide votes are never distributed across counties. A source export covering every county is not proof that counting has finished.
Provisional 2026 review-border rules
Each rule below can independently trigger a review border from a county’s own historical change. Democratic and Republican rules, gains and losses are symmetric. State-peer medians and modified-z scores remain context and cannot veto these own-history rules. These thresholds are application choices, not probability estimates.
- Party counts: absolute change of at least 40%, at least 100 votes, while the opposing party changes by at most 15% in absolute size. Require at least 500 two-party votes in each year and 100 votes for each prior party.
- Reported race votes: absolute change of at least 40% and 100 votes, with at least 500 candidate/write-in race votes in each year. Overvotes, undervotes, invalid/blank/spoiled/exhausted categories and reported ballot totals are excluded. Unknown or unitemized categories withhold the total.
- Actual ballots per resident: absolute change of at least 10 percentage points, at least 100 explicitly reported ballots and 1,000 residents each year. Current 2026 ballots use exact 2025 population; the final/certified 2022 ballot baseline uses exact 2022 population.
- Race votes per resident: a separate proxy with an absolute change of at least 10pp, using qualified race-vote totals and exact population reference years. This proxy is never labeled actual ballots or eligible-voter turnout.
- Two-party share: an absolute Democratic/Republican share swing of at least 10pp, with the party minimums above. Other votes stay unassigned.
Residents include children and other people ineligible to vote. No ACS denominator is used. A 100% precinct-reporting figure does not establish complete ballot counting or certification. A missing source never generates a border merely for being missing.
Baseline years and qualified scopes
Governor compares 2022 with 2026, including two-year-term states; it is not labeled the immediately previous election. Senate compares the same regular Class II seat in 2020 and 2026 only where unique general, non-special records and source quality are established. The 2020 presidential-election cycle differs from the 2026 midterm, so broad race-vote changes can reflect normal cycle differences. Georgia/Louisiana first-round ambiguities and unqualified special elections are withheld.
House comparisons currently qualify only Montana’s 57 verified portions across 56 counties and Arizona’s 30 portions across 15 counties/all 9 districts in both 2022 and 2026. Arizona’s 2022 source is a saved primary certified-PDF text extraction, not downloaded PDF bytes. Counting completeness is not inferred from the complete 2026 scope. All other House comparison scopes remain unassessed until separately qualified.
Assessment coverage in this snapshot
| Screen | Office | Assessed counties | Not assessed | Flagged counties |
|---|---|---|---|---|
| Party count change | U.S. Senate | 647 | 2,495 | 26 |
| Party count change | U.S. House | 57 | 3,085 | 17 |
| Party count change | Governor | 1,051 | 2,091 | 290 |
| Reported race-vote change | U.S. Senate | 873 | 2,269 | 11 |
| Reported race-vote change | U.S. House | 67 | 3,075 | 0 |
| Reported race-vote change | Governor | 1,080 | 2,062 | 50 |
| Race votes per resident | U.S. Senate | 870 | 2,272 | 654 |
| Race votes per resident | U.S. House | 67 | 3,075 | 0 |
| Race votes per resident | Governor | 1,078 | 2,064 | 76 |
| Two-party share swing | U.S. Senate | 647 | 2,495 | 89 |
| Two-party share swing | U.S. House | 57 | 3,085 | 9 |
| Two-party share swing | Governor | 1,051 | 2,091 | 295 |
Actual-ballot/resident screen: 58 assessed, 3,084 not assessed, 22 flagged. The audit assigns a status to all 3,142 atlas counties, which does not mean all are assessed. One county can trigger multiple rules; flagged counts must not be summed as unique findings.
What a flag means
A flag identifies a reported change for source review. Candidate changes, reporting progress, election cycles and geographic composition may explain it. It is not a finding that votes are wrong and does not identify a cause. Signed party directions follow the county’s actual change; neutral total/ratio flags do not acquire a party color, and conflicting directional evidence remains mixed.
Source update times, retrieval times and snapshot times remain distinct. Unknown count-update times stay unknown. The full pattern-review report and the machine-readable audit retain exact evidence, thresholds, source links, input hashes and non-assessment reasons.
Primary and research references
- MEDSL County Presidential Election Returns 2000–2024 (via Election Data mirror)
- MEDSL U.S. Senate statewide 1976–2024
- MEDSL U.S. House 1976–2024 (via Election Data mirror)
- Census Vintage 2025 county population estimates
- U.S. Senate Class II seat/term record
- NIST exploratory modified-z reference — peer context only
See data source documentation and the source-specific guides for exact dataset versions and limits. Upstream data and dependency licenses remain separate from the application’s MIT license.