2026 at a glance
Senate map
Each square is a state, colored by the Democratic win chance in its 2026 Senate race (Estimate, from this site's model). Gray = no Senate race modeled. Tap a state to open its race.
Closest races
Trendlines
Wisconsin
Wisconsin has no 2026 U.S. Senate race; its open governor race (Evers retiring) is listed below with the House races. Its data is thinner than Texas: no county-by-county official daily feed is wired in here, so this tab shows the modeled House races and the statewide early-vote row from the UF Election Lab aggregator.
As of 10/30/2026 · original source: https://elections.wi.gov/resources/statistics/absentee-ballot-report-november-3-2026-general-election-0 · via UF Election Lab
| Measure | All | Dem | Rep | None/other |
|---|---|---|---|---|
| Mail/early ballots requested | 787,852 | — | — | — |
| Mail ballots returned/accepted | 299,709 | — | — | — |
| In-person early votes | 393,625 | — | — | — |
| Total early votes | 693,334 | — | — | — |
2022 general-election total turnout: 2,673,154.
Voter registration change since 2022 / 2024
Not yet compiled for this state.
This state does not register voters by party. Official registration counts: the state election office, or the U.S. EAC Election Administration and Voting Survey. No figure is shown until it is pulled from an official report.
Official absentee / early vote
Wisconsin Elections Commission (official): WEC absentee ballot report, Nov 3, 2026 general election. Wisconsin's in-person early voting counts as absentee voting, so these reports cover ballots requested, sent and returned. Link only. WEC blocks automated requests, so the figures in the table above come from UF Election Lab's copy of the reports.
Other sources
Wisconsin Elects secondary · independent (announced by @TobyMGData). It is an independent site by Joshua Rifelj and is not affiliated with WEC. It plans to post election-night precinct-level maps for competitive Wisconsin State Assembly and Senate districts. It does not cover early or absentee vote and has no data download or API, so this page only links to it. As of Oct 20 the site was marked "coming soon".
Most watched
Closest races by projected margin (Estimate), then the most-polled races (polls in the last 60 days). Tap a row for detail.
Most competitive
Most polled
All states · Senate
Every 2026 Senate race this model covers, closest first. Tap a row for detail.
All states · House
Every House race modeled race by race, closest first. Remaining safe seats are counted in the House total but not listed.
All states · Governor
All 36 governor races on Nov 3, 2026, closest first. Same pipeline as the Senate: graded polls (excluded firms removed), 12-year state poll-miss correction (cap ±3), CNN Trump-state offset (Senate-derived, max() target, not double counted), ratings, 2024 presidential base + national environment, incumbency and party control. Texas also uses its statewide early-vote projection. Each row lists the incumbent status, state trend, ticket-splitting history and the governor’s over/under (display only). Tap a row for detail.
Texas governor
Abbott (R, incumbent) vs Hinojosa (D). Same card as All states Governor.
Who's showing up: R vs D
Through Day 10, voters with primary history still lean Democratic, but barely (D+0.9, a 56,920-vote edge), unlike the same point in 2022 (R+12.3, R+524,853) and 2024 (R+11.9, R+896,695). The edge shrank from D+9.7 on Day 1 to D+2.1 on Day 5, ticked up to D+2.6 after the weekend, then narrowed to D+2.0 (Day 8), D+1.2 (Day 9) and D+0.9 (Day 10). Republicans still have about 323K more past primary voters who have not voted yet. Too early to call.
How this is measuredOutcome projection: statewide R−D margin
Your assumptions, your model. Move a slider and the projection updates; settings stay on this phone. The single biggest unknown is how voters with no primary history break. That one setting moves the result more than everything else combined.
Backtest (honest errors)
| Test | Actual | Model (defaults) | Miss | |
|---|---|---|---|---|
| 2018 final (Senate) | R+2.6 | R+12.5 | +9.9 R | FAIL |
| 2022 final (Gov) | R+10.9 | R+9.7 | −1.2 | pass |
| 2022 Day 1 → final | R+10.9 | R+10.7 | −0.2 | pass |
| 2024 Day 1 → final | R+13.7 | R+10.0 | −3.7 | FAIL (>3) |
Why this differs from the “Who's showing up” card: both use Ryan's Day 3 data, but they are different models by design. That card shifts past final margins by half the change in primary-history lead (R+2.9). This projection adds the not-yet-voted primary voters and independents to the Day 3 electorate, so it gives a different number. With the same Day 3 inputs, both are ESTIMATES.
Same model, independent split fixed at 0.50, composition from Ryan finals / Day 1 (backtests) and Day 3 (2026 inputs) reports Verified inputs. Two of four tests miss by more than 3 points, so this is a scenario tool, not a forecast, and confidence is capped at Low. A tested signal (new-voter share → independent split) made errors worse (2018 miss 16.5, 2022 Day 1 4.3) and was dropped. No 2024 final composition and no 2018/2022 pool tables are on file, so the headroom step is untested Unverified.
If trends hold: 2026 Senate & House
What-if, not a forecast. Confidence: Low. Actual vote data comes first where it exists: in Texas the early-vote projection carries 70% of the weight, rising as more of the expected vote is banked. Next come race polls, with fake firms and outlier polls removed (see "Excluded inputs"), then a poll-bias correction (the average past error for midterms, per state, adjustable below) [v3.1, Oct 23 2026: the midterm per-state term is now a 12-year state poll miss: each state's Senate, governor and presidential polls from 2014–2022 (final 21 days, FiveThirtyEight raw polls) vs results, shrunk toward zero, capped at ±3 pts. Held-out backtest on Senate 2016–2022: typical miss 5.27→4.95 pts (better), but winners called 114→110 of 124 (worse). Kody chose the smaller typical miss.], then the race ratings, and then the national-environment prior built on each race's 2024 presidential margin, adjusted for incumbency, money and a small party-control history term. North Carolina's early vote is calibrated against the 2022 same-point file and propensity-weighted: who has voted vs. who is left, by vote history and party (NCSBE voter file; 75% cap). Texas uses primary-history propensity tiers. Then an undecided-allocation term. Other states' early-vote counts are shown for context only (weight 0) until a 2022 same-point baseline is found.
Senate (of 100 · 51 control)
VP JD Vance (R) breaks 50–50 ties, so 50 D = R control. Not up: D 34 (incl. 2 I), R 31 (Verified).House (of 435 · 218 control)
79 contested seats (Cook Toss-up/Lean/Likely) modeled individually; 356 safe seats use baseline + national pressure only.Senate races
All other Senate races
Closest House seats
All 79 contested House seats
Excluded inputs (33 polls)
Removed before averaging: exposed/fabricated or banned firms (Wedgewood, Median Strategies, Public Sentiment Institute, Research 2000, Strategic Vision, Silver Bulletin banned list, FiftyPlusOne-suspended Patriot Polling, SoCal Strategies / SoCal Research — linked by FiftyPlusOne to Wedgewood, site offline Oct 2026; can be restored if FiftyPlusOne clears it), non-scientific panels, n<300, and outliers more than 2.5 scaled MADs from the race median after the house-effect adjustment (races with 4+ polls). Polls with no published n are kept (gap). No row could be identified as unweighted opt-in from table metadata.
CNN Trump-state poll offset (added Oct 23)
Source: CNN chief data analyst Harry Enten, CNN News Central, Oct 21, 2026 (transcript): in 2018–2024 Senate races, Republicans beat their final poll averages by 4.1 pts in states Trump won vs 0.2 pts in states he lost. How the tool uses it: for Senate races with polls, the total poll-miss correction is the larger of the existing 12-yr state correction (× strength) and the CNN target (R+4.1 in 2024 Trump states, R+0.2 elsewhere). Only the missing part is added, so nothing is double counted. The ±3 cap still applies to the 12-yr state figure; the CNN target may reach 4.1. House: not applied, because House seats have no race polls and the poll-miss correction never flows to them. Added at Kody’s direction (Oct 23) as a published, verifiable offset; it did not pass the out-of-sample test. Context (held-out Senate 2016–22, 123 races, Estimate): a Trump-state offset learned from prior cycles cut typical miss in presidential years (2016 5.34→3.96, 2020 6.69→5.07) but overshot toward D in midterms (2018 3.84→4.46, 2022 4.32→4.83). The flat CNN 4.1/0.2 scored 4.17 vs 5.06 overall but is in-sample (built from those years).
Age-weighted electorate tilt (added Oct 23)
Applied by Kody’s direction (Oct 23); not backtest-adopted. What it does: re-weights the expected electorate toward voters 50–75 and away from under-30s, then adds the resulting margin shift once to every Senate, House and Governor projection (polls and non-poll inputs alike). Weights (Kody-directed choices, not sourced): 18–29 ×0.90; 50–64 ×1.05; 65–74 ×1.05; 30–49 and 75+ unchanged (65+ band treated as 60% aged 65–74, so ×1.03). Sources: electorate shares and party splits from the CNN/Edison 2022 national House exit poll (CNN): 18–24 7% D+25, 25–29 6% D+32, 30–39 14% D+11, 40–49 14% R+7, 50–64 31% R+11, 65+ 28% R+12; 65–74 share of 65+ voters (18.4 of 30.5 pts) from Census CPS Nov 2022 (P20-586). Result: national margin moves from R+2.54 to R+3.14, a shift of R+0.6 (Estimate). Since the EV year-type proration below, each race gets R+0.6 × its state’s w. Double counting: most polls already weight by age, so the tilt is kept small (half-point scale) and applied once at the end; it is separate from and does not change the CNN max(), the 12-yr poll-miss correction or the House coattail (coattails read raw Senate/Governor polls). Backtest (Senate 2016–22, 123 races, Estimate): on top of the current model typical miss 4.24→4.25, winners 111→111; better in presidential years (2016 4.03→3.86, 2020 4.95→4.64), worse in midterms (2018 3.79→4.10, 2022 4.22→4.45) and in close races (3.38→3.51). So the data do not support it for a midterm; it reflects Kody’s judgment.
EV year-type proration (added Oct 23)
Applied by Kody’s direction (Oct 23): 2026 is not treated purely as a midterm. Each midterm-vs-presidential factor = w × presidential-year value + (1−w) × midterm value, where w (0–1, share presidential-like) comes from 2026 early-vote turnout versus the same point in 2022 and 2024: w = (2026 − 2022) / (2024 − 2022). Measured same-point inputs (updated Oct 29 evening; 2026 data through Oct 23 to 28, matched to the same early-vote day where start dates differ, otherwise the same days before Election Day): North Carolina w 0.27 (NCSBE absentee file: 2,141,660 accepted through EV day 14 vs 1,585,590 in 2022 and 3,648,485 in 2024; Verified source). Texas w 0.54 (TX SOS, all 254 counties, EV day 9 revised, share of registered voters: 29.91% vs 21.52% in 2022 and 37.12% in 2024; Verified source; Day 10 skipped because the Harris Wednesday count is partial). Georgia w 0.24 (GA SOS absentee voter file, EV day 12: 1,751,150 vs 1,426,704 and 2,791,300; raw counts; Estimate; not refreshed Oct 26 to 29, GA SOS file page needs a browser; 2022 day 12 was a Friday). Michigan w 0.27 (MI SOS release, 7 days out: 1,392,170 vs 1,126,246 in 2022 and 2,108,478 in 2024; Estimate; distorted: permanent mail list auto-mails ballots since 2023 and in-person early voting is new since 2024; mail only would give 0.10; moved more than 0.15 from a stale value). Kansas w 0.45 (KS SOS via IKE Lab, 7 days out: 245,001 vs 167,620 and 340,048; Estimate). Maine w 0.58 (ME SOS absentee file, 8 days out: 212,349 vs 153,216 and 254,444; Estimate). Iowa w 0.27 (IA SOS daily absentee file, 11 days out: 207,858 vs 167,105 and 317,952; Estimate; not refreshed since Oct 25, IA SOS site blocked automated access). Michigan, Kansas and Maine were not refreshed Oct 28 or 29 (values from Oct 27). Not measured, fallback kept: Ohio (no official 2022 and 2024 figures at the same point; also the 2026 end of the mail grace period pushes returns earlier); Florida (2023 law expired standing mail requests, so 2024 mail at the same point was below 2022 and the ratio is undefined; no same-point in-person history); New York (early voting starts Oct 24; measured from then). Daily series compiled from the official files by the sethjlevyhub early-vote tracker (labeled aggregator) and checked against SOS releases where available. Other states: the projected-final-turnout position (projected 2026 total between 2022 and 2024 totals, the Turnout-scaled strength t); most of those projections are prorated from NC, so they are weaker (Estimate). National w = 2022-turnout-weighted mean of state w: …. w updates as early-vote data updates. Party, age and new-voter composition are not yet used in w (not available same-point for 2022 and 2024 in the tool). Blended factors, replacing (not stacking on) the year-type-agnostic values: (1) CNN Trump-state target, R pts: Trump-won states 4.1 + 1.02w − 1.26(1−w); other states 0.2 + 3.17w − 2.95(1−w). These are CNN’s 4.1/0.2 plus the year-type deviation in raw Senate poll miss 2016–22 (123 races): Trump states presidential R+6.23, midterm R+3.95, all R+5.21; other states presidential R+3.23, midterm D+2.89, all D+0.06. Still a max() against the 12-yr state correction, so only the missing part is added. (2) Age tilt: R+0.6 × w (helped 2016/2020, hurt 2018/2022, so the midterm value is 0). Not split by year type: the 12-yr state poll miss (too few races per state per type), the House coattail cap. Backtest (Senate 2016–22, leave-one-cycle-out year-type values, w = national VEP turnout position 2016 1.0, 2018 0.57, 2020 1.0, 2022 0.0; Estimate): typical miss 4.25→4.19, winners 111→113, close races 3.51→3.40; 2016 3.86→4.53 worse, 2018 4.10→3.90, 2020 4.64→4.06, 2022 4.45→4.23 better. Only four cycles, so treat as indicative.
Backtest: 2018 & 2022 Senate
Prior + FPS-weighted polls vs. a uniform swing, with the actual House popular vote as the national environment. 2018 (30 races): mean error 3.67 vs 8.44 pts, winners 27 vs 26. 2022 (30 races): 4.45 vs 6.9 pts, winners 26 vs 25. Caveats: the national environment is known in hindsight, so 2026 will do worse. FPS was partly built on these years (leakage). Incumbency, ratings and money weren't tested. House not backtested. v2.7 leave-one-year-out, Senate 2016–2022 (538 raw polls; 2024 polls not available), mean error / winners right of 114: baseline 4.51 / 105; + flat incumbency 4.138 / 103; incumbent-specific 4.205 (worse, dropped); + same-type bias correction 3.638 / 107; + party history 3.62 / 107 (kept). Bias correction mixing all years 4.358 (worse than same-type).
What else we account for (19 factors)
Drawn from the biggest 2016–2024 poll misses (AAPOR 2016 & 2020 task forces, Pew 2021). Each measurable factor was tested leave-one-year-out on 114 Senate races 2016–22: shown as mean error / winners right, against a base of 3.62 / 107. A factor was kept only if error fell without losing a winner.
Method & sources
Prior = 2024 pres margin + elasticity (1; Unknown: no public per-seat elasticity loaded; all seats 1.0) × national swing + incumbency (6 pts; Jacobson (2015), J. Politics 77(3): House incumbency worth 3.0–3.7 pts of vote share in 2014; lower bound 3.0 share ≈ 6.0 margin pts. Applied to Senate as an assumption.) + money (clamp(1.5·log10(D receipts / R receipts), ±1.5) added to prior; Assumption, untested). Prior SD 7. Evidence: Cook/IE/Sabato ratings mapped to margins (Toss-up 0, Tilt 1.5, Lean 4.5, Likely 9, Solid 16; SD 6). Polls from the last 60 days, weighted by FPS score × recency (half-life 21 d), RV ×0.8, house bias ×0.5 removed; SD sqrt(5²/n_eff + 3²). Texas races also use the early-vote projection (SD 8), and it gets 10% of the national swing. Combined by precision weighting. Age tilt: R+0.6 × w per race (50–75 weighted up, under-30 down), applied by Kody’s direction. Year-type proration: midterm-vs-presidential factors are blended by each state’s early-vote similarity w (see EV year-type proration). 4000 simulations with a shared national error (SD 2.5) give totals and 10th–90th ranges. Independents count on the anti-R side. Alaska ranked-choice voting isn't modeled. The House baseline is presidential vote by 2026 district.
- Senate seats up & not up; Cook ratings Oct 20 2026 (House Oct 23); generic-ballot aggregates
- 2024 presidential results by state
- 2024 presidential by 2026 House district (The Downballot)
- Senate race polls: Wikipedia 2026 Senate election pages for each state (pulled 2026-10-22)
- House ratings (Cook Oct 23, IE Oct 15, Sabato Oct 20)
- FEC candidate totals (through 2026-09-30)
- Incumbency advantage: Jacobson 2015
- RealClearPolling generic ballot
- Silver Bulletin
Confidence report
One row per race. Confidence (Estimate, rule of thumb): High = win chance at least 85/15 with 10+ polls (or 95/5); Medium = at least 70/30 with 3+ polls (or 85/15); else Low. Key drivers = largest terms in the adjustment stack. Tap a column header to sort. Projections are unchanged by this view.
Loading…
Statewide and county totals: TX SOS EV rosters (Day 10, Wed Oct 28; Harris Wednesday count appears partial, about 51K below Ryan). Pace: TX SOS Day 9 (revised to 5,695,755) vs the same EV day in 2022 (official archive). Primary-history figures: Ryan Data & Research Day 10 (emailed Oct 29). Live SOS numbers are in “Official data” below.
Charts
100% = same turnout as 2022 at the same point. County figures from county election offices (via local reporting) Verified; same-day 2024 county pace: awaiting data.
Looks like vs. actually means
Looks likeRecord turnout = a blue wave.
Actually meansMore than 90% of Day 1 voters also voted in 2024; most are habitual voters voting earlier. In 2018, hot early turnout still ended R+2.6.
Looks likeDemocrats lead the early vote by primary history.
Actually meansDemocrats are front-loading: through Day 10, 87% of 4-time D primary voters have voted vs 80% of 4-time R. Republicans have about 323K more primary voters who have not voted yet (1,480,278 vs 1,157,346; Ryan Day 10).
Looks likeMail ballots are mostly Democratic.
Actually meansTrue now (50% D / 30% R history), a flip from 2010. Mail is only ~5% of all votes.
Looks likeThe state website shows low numbers in Harris and Fort Bend, and high in Dallas.
Actually meansThe SOS feed lags or misreports county uploads. Through Day 3 it is ~69K too high in Dallas and ~70K/44K too low in Harris/Fort Bend. County offices are the better count.
Looks likeRed counties like Montgomery are outpacing 2018 more than Dallas or Travis.
Actually meansRaw ratios ignore growth: Montgomery has ~49% more registered voters than in 2018, Dallas ~16%. Per registered voter the order flips (Dallas ~98%, Travis ~91%, Montgomery ~80% of 2018 pace; unofficial partial-day data).
Polls: who to trust
Fairfield Pollster Score (0–100)
Our composite, sorted by score. Silver's grade stays visible. Estimate Weights: Silver Bulletin predictive +/− 30% · 538 POLLSCORE (2024 ratings archive) 20% · our measured accuracy 30% · transparency 20%. Missing input = n/a, and the weight is re-spread across the rest (nothing imputed). Penalties: herding (−10 × Silver's herding penalty); banned or fabricated = 0 and not listed. Our accuracy = each firm's final-3-week general-election polls 2018–2024 (Silver Bulletin raw poll archive, 12,350 polls), error vs. other firms in the same race. ↕ = our rank differs from Silver's by 10+ places. Backtest: no reliable edge over Silver's grade (see method).
| # | Pollster | FPS | Silver | 538 | Our err (bias) | Polls | TX | Conf |
|---|---|---|---|---|---|---|---|---|
| 1 | AtlasIntel | 92 | A | 2.7 | 1.7 (D+0.2) | 83 | 3 | High |
| 2 | The Washington Post | 91 | A+ | 3 | 3.1 (D+2.8) | 21 | 0 | Medium |
| 3 | Suffolk University ↕ | 90 | A- | 2.9 | 2.5 (D+0.8) | 21 | 0 | Medium |
| 4 | Emerson College | 89 | A | 2.9 | 3.8 (D+0.7) | 202 | 7 | High |
| 5 | The New York Times/Siena College | 88 | A | 3 | 3.9 (D+1.5) | 104 | 4 | High |
| 6 | Data Orbital | 87 | A/B | 2.9 | 3 (R+0.3) | 10 | 0 | Low |
| 7 | Beacon Research/Shaw & Co. Research | 87 | A- | 2.8 | 3.4 (D+2.1) | 28 | 0 | Medium |
| 8 | Hart Research Associates/Public Opinion Strategies | 82 | A- | 2.6 | 2.6 (D+1.9) | 10 | 0 | Low |
| 9 | University of Massachusetts Lowell ↕ | 82 | A- | 2.9 | 4.1 (D+3.6) | 20 | 2 | Medium |
| 10 | Marist College | 81 | A- | 2.9 | 4.1 (D+3.5) | 44 | 0 | Medium |
| 11 | Research & Polling Inc. | 78 | A+ | 2.1 | 3.9 (R+0.3) | 13 | 0 | Low |
| 12 | Landmark Communications | 77 | A- | 2.1 | 2.6 (D+0.1) | 12 | 0 | Low |
| 13 | East Carolina University ↕ | 77 | B/C | 2.6 | 3.5 (D+3.1) | 15 | 0 | Low |
| 14 | Quinnipiac University | 76 | B+ | 2.8 | 4.9 (D+4.2) | 30 | 4 | Medium |
| 15 | Saint Anselm College ↕ | 76 | A/B | 2.4 | 4.2 (R+0.7) | 13 | 0 | Low |
| 16 | Cygnal ↕ | 74 | A | 2.1 | 3.9 (D+2.7) | 53 | 3 | High |
| 17 | Echelon Insights | 74 | A- | 2.7 | 2.2 (R+0.2) | 17 | 0 | Low |
| 18 | Monmouth University ↕ | 73 | A | 2.9 | 6.1 (D+4.6) | 25 | 0 | Medium |
| 19 | TIPP Insights ↕ | 72 | A | 1.8 | 1.8 (D+1.7) | 27 | 0 | Medium |
| 20 | Siena College | 71 | A | 2.7 | 6.3 (D+1.9) | 17 | 1 | Low |
All 58 scored firms (10+ polls)
| # | Pollster | FPS | Silver | S/I/O/T | Pen |
|---|---|---|---|---|---|
| 1 | AtlasIntel | 92 | A | 98/96/93/82 | 1 |
| 2 | The Washington Post | 91 | A+ | 100/100/80/96 | 2.2 |
| 3 | Suffolk University ↕ | 90 | A- | 96/98/84/93 | 2.3 |
| 4 | Emerson College | 89 | A | 98/99/83/91 | 3.1 |
| 5 | The New York Times/Siena College | 88 | A | 99/100/74/94 | 2.8 |
| 6 | Data Orbital | 87 | A/B | 99/97/76/95 | 4 |
| 7 | Beacon Research/Shaw & Co. Research | 87 | A- | 97/99/75/84 | 1 |
| 8 | Hart Research Associates/Public Opinion Strategies | 82 | A- | 96/95/78/79 | 4.9 |
| 9 | University of Massachusetts Lowell ↕ | 82 | A- | 93/98/57/94 | 1.5 |
| 10 | Marist College | 81 | A- | 95/97/51/96 | 0.8 |
| 11 | Research & Polling Inc. | 78 | A+ | 100/96/97/14 | 2.9 |
| 12 | Landmark Communications | 77 | A- | 95/94/96/16 | 2 |
| 13 | East Carolina University ↕ | 77 | B/C | 80/91/73/88 | 5.3 |
| 14 | Quinnipiac University | 76 | B+ | 90/93/36/98 | 0 |
| 15 | Saint Anselm College ↕ | 76 | A/B | 87/91/99/22 | 2.6 |
| 16 | Cygnal ↕ | 74 | A | 99/99/91/13 | 5.3 |
| 17 | Echelon Insights | 74 | A- | 97/96/81/44 | 7.3 |
| 18 | Monmouth University ↕ | 73 | A | 98/97/26/100 | 3.6 |
| 19 | TIPP Insights ↕ | 72 | A | 99/89/88/12 | 4.8 |
| 20 | Siena College | 71 | A | 97/96/22/87 | 1.9 |
| 21 | OH Predictive Insights | 70 | B+ | 91/88/82/21 | 4.1 |
| 22 | University of New Hampshire ↕ | 68 | B- | 48/82/63/99 | 1.9 |
| 23 | Ipsos ↕ | 67 | B | 76/97/42/88 | 4.9 |
| 24 | SurveyUSA ↕ | 67 | A | 98/100/44/29 | 1.3 |
| 25 | Selzer | 66 | B+ | 89/99/7/86 | 0 |
| 26 | InsiderAdvantage | 62 | B+ | 84/83/61/16 | 0.9 |
| 27 | Susquehanna Polling & Research Inc. ↕ | 61 | A- | 95/82/43/20 | 0.6 |
| 28 | YouGov | 61 | B | 78/98/32/45 | 1.2 |
| 29 | Gravis Marketing/Kaplan Strategies ↕ | 60 | B- | 69/97/59/12 | 0.5 |
| 30 | Alaska Survey Research | 60 | A/B | 91/73/47/18 | 0 |
| 31 | OnMessage Inc. ↕ | 60 | A | 99/27/94/8 | 5.2 |
| 32 | RMG Research | 60 | B | 76/85/62/20 | 2.7 |
| 33 | Civiqs | 57 | B | 77/76/35/43 | 0 |
| 34 | Rasmussen Reports | 56 | B+ | 84/n/a/67/0 | 0.9 |
| 35 | KAConsulting | 56 | A/B | 87/70/66/12 | 5.9 |
| 36 | Research Co. | 55 | B | 81/92/69/21 | 12.4 |
| 37 | co/efficient | 52 | B | 76/14/86/16 | 3 |
| 38 | University of Southern California | 51 | B/C | 51/77/9/92 | 0 |
| 39 | Harris Insights & Analytics | 50 | B- | 61/73/49/14 | 0.2 |
| 40 | Remington Research Group | 50 | B- | 50/94/37/24 | 0 |
| 41 | Wick | 50 | B/C | 77/39/52/16 | 0 |
| 42 | St. Pete Polls | 50 | C+ | 31/82/64/26 | 0 |
| 43 | Public Policy Polling ↕ | 48 | B+ | 83/58/28/14 | 0 |
| 44 | Redfield & Wilton Strategies | 48 | B | 76/9/34/96 | 6.3 |
| 45 | Data for Progress | 45 | C+ | 32/99/45/20 | 2.1 |
| 46 | Swayable | 44 | C+ | 30/50/31/85 | 0.9 |
| 47 | Targoz Market Research | 43 | B- | 54/36/18/73 | 0.3 |
| 48 | SSRS | 41 | C+ | 16/77/20/94 | 4.6 |
| 49 | Trafalgar Group ↕ | 41 | B | 77/5/53/6 | 0 |
| 50 | Florida Atlantic University/Mainstreet Research ↕ | 40 | B+ | 88/n/a/39/0 | 7.1 |
| 51 | Mitchell Research & Communications | 38 | C | 6/78/54/35 | 2.3 |
| 52 | SurveyMonkey | 38 | D+ | 2/25/79/45 | 0 |
| 53 | Morning Consult | 36 | C+ | 21/83/33/16 | 0 |
| 54 | GBAO ↕ | 34 | B/C | 75/43/14/6 | 2.6 |
| 55 | Patriot Polling ↕ | 34 | B | 74/4/55/18 | 8.7 |
| 56 | Change Research | 33 | C+ | 20/6/40/78 | 2.1 |
| 57 | Citizen Data | 9 | B/C | 9/7/10/14 | 0.4 |
| 58 | The Political Matrix/The Listener Group | 5 | C/D | 2/3/8/9 | 0 |
538 grade is 0–3 (3 = best). Bias: “D+1.0” = overstated Democrats. Confidence: High = all 4 inputs and 50+ polls; Medium = 3+ inputs and 20+ polls; else Low.
Silver Bulletin top-rated pollsters (reference)
Source: Silver Bulletin pollster ratings, updated Jan 14, 2026 (full data file) Verified. Rank among all rated firms; this list shows firms with 20+ rated polls. Plus-minus: negative = more accurate than average. Bias = historical lean, mean-reverted; “D+0.9” means it overrated Democrats by 0.9 on average. TI = AAPOR Transparency Initiative or Roper archive (AAPOR TI).
| # | Pollster | Grade | +/− | Bias | Polls | TI |
|---|---|---|---|---|---|---|
| 1 | The Washington Post | A+ | -1.09 | D+0.9 | 98 | ✓ |
| 3 | Research & Polling Inc. | A+ | -0.98 | R+0.6 | 44 | |
| 4 | OnMessage Inc. | A | -0.85 | R+1.2 | 32 | |
| 6 | TIPP Insights | A | -0.81 | R+0.1 | 51 | |
| 7 | Cygnal | A | -0.80 | D+2.0 | 58 | |
| 8 | NYT/Siena College | A | -0.80 | D+1.4 | 116 | ✓ |
| 9 | Monmouth University | A | -0.77 | D+2.0 | 131 | ✓ |
| 10 | SurveyUSA | A | -0.66 | D+0.1 | 871 | |
| 11 | CBS News/NYT | A | -0.65 | D+0.3 | 66 | ✓ |
| 12 | Public Opinion Strategies | A | -0.64 | R+1.3 | 80 | ✓ |
| 13 | AtlasIntel | A | -0.64 | R+0.0 | 93 | ✓ |
| 14 | Emerson College | A | -0.61 | D+0.6 | 320 | ✓ |
Texas-relevant pollsters
| # | Pollster | Grade | Bias | Polls | Note |
|---|---|---|---|---|---|
| 14 | Emerson College | A | D+0.6 | 320 | Polls TX often (2026: Oct 18–20) |
| 58 | Quinnipiac University | B+ | D+0.6 | 229 | |
| 88 | InsiderAdvantage | B+ | D+0.3 | 208 | R-sponsored TX polls in 2026 |
| 89 | Rasmussen Reports | B+ | R+1.4 | 797 | |
| 118 | YouGov | B | D+1.1 | 521 | Fields CBS & UT/TPP Texas polls |
| 127 | Trafalgar Group | B | R+1.9 | 143 | |
| 257 | University of Houston (Hobby School) | B/C | n/a | 1 | Sparse data: 1 rated poll |
| 429 | UT Tyler | B/C | D+0.8 | 10 |
UT / Texas Politics Project and Texas Lyceum: not found in Silver Bulletin's database (no grade) Unverified. How this feeds the index: C5 (polls) weights by these grades, house-effect-adjusts by the bias column, and excludes F/banned firms and ActiVote-style unscientific polls. C5 is still not scored (no graded TX average compiled yet).
Polls reported as fake or not real
Only cases documented by news outlets, aggregators or AAPOR. Wording follows the source.
Confirmed or admitted fabrication
- Research 2000 2010-06-29
Daily Kos (client) said an independent statistical review found its weekly polls could not be random samples and sued; the case settled in 2011 without a ruling on fabrication. Daily Kos - Median Strategies 2026-08-18
Released an L.A. mayor poll later admitted to be fake, described by its creator as a “social experiment.” LA Times - The Public Sentiment Institute 2026-08-20
Admitted fabricating vote-choice data for some respondents (Florida GOP governor primary); permanently banned by FiftyPlusOne. FiftyPlusOne - Wedgewood Polls 2026-10-20
Publicly admitted after a FiftyPlusOne inquiry that its midterm polls were fabricated; aggregators removed its data. FiftyPlusOne · Guardian
Flagged, questioned, suspended or banned (no fabrication finding)
- Strategic Vision LLC 2009-09-23
AAPOR publicly rebuked it for refusing to disclose basic methodology for 2008 primary polls. A client (OCPA) later pulled a survey it could not verify. Fabrication was alleged by critics, not proven. OCPA - Patriot Polling 2026-08-20
Suspended by FiftyPlusOne pending answers on alleged undisclosed campaign sponsorship. No fabrication finding. FiftyPlusOne - Silver Bulletin banned list 2026-01-14
Grades these as F (banned): TCJ Research, Pharos Research Group, Big Data Poll, Overtime Politics, KG Polling, and others. Reasons are not itemized in the table. Silver Bulletin - ActiVote 2026-01-14
Not rated by Silver Bulletin: not treated as a scientific survey (convenience sample). Not fake; treat with caution. Silver Bulletin
Latest polling news (11 links)
- Pollwatcher: another fraudulent pollster out of business; a third group vanished
2026-10-22 · Washington Post - Texas poll (Oct 17–19, n=1,000 LV): Senate a toss-up, Talarico 48–46 before leaners
2026-10-20 · Emerson College - Wedgewood Polls admits fabricated data, calls it a “social experiment”
2026-10-21 · The Guardian - Fox News poll: Talarico and Paxton tied at 49% in the Texas Senate race
2026-10-21 · Fox News - How the Wedgewood fabrication was caught (vetting details)
2026-10-20 · FiftyPlusOne - Pew warning: 2026 midterms will be uniquely hard for polls (turnout)
2026-10-14 · Washington Post - Texas Senate polling average (quality- and house-effect-adjusted)
2026-10-22 · FiftyPlusOne - Why it banned TPSI and suspended Patriot Polling
2026-08-20 · FiftyPlusOne - Site behind widely shared fake polls (Median Strategies) run by a 21-year-old
2026-08-20 · The Guardian - 2025 report card: sometimes polls underestimate Democrats (Trafalgar R-bias)
2026-01-14 · Silver Bulletin - Full pollster ratings, updated after 2025 (incl. herding penalty)
2026-01-14 · Silver Bulletin
Links checked Oct 22, 2026. Refreshed nightly (see README).
Official data: SOS feed tables (live)
Page generated Oct 30, 5:34 PM CT.
Texas statewide (cumulative)
Source: Texas Secretary of State, unofficial early voting turnout (data as of report date 10/30/2026, SOS file updated 2026-10-30). 254 of 254 counties show in-person votes so far. Counties upload voter rosters on their own schedule, so the SOS feed can lag county-announced totals by a day or more (e.g., a county may announce its own count before its roster appears here).
By day
| Report date | In person that day | Cum. in person | Cum. mail | Cum. total | % of reg. |
|---|---|---|---|---|---|
| 10/14/2026 | 0 | 768 | 78,749 | 79,517 | 0.4% |
| 10/15/2026 | 0 | 768 | 92,826 | 93,594 | 0.5% |
| 10/16/2026 | 3 | 771 | 105,031 | 105,802 | 0.6% |
| 10/17/2026 | 0 | 771 | 108,683 | 109,454 | 0.6% |
| 10/18/2026 | 0 | 771 | 109,451 | 110,222 | 0.6% |
| 10/19/2026 | 784,599 | 785,370 | 125,680 | 911,050 | 4.8% |
| 10/20/2026 | 745,512 | 1,530,882 | 139,485 | 1,670,367 | 8.8% |
| 10/21/2026 | 684,247 | 2,215,129 | 156,776 | 2,371,905 | 12.5% |
| 10/22/2026 | 633,713 | 2,848,842 | 171,437 | 3,020,279 | 15.9% |
| 10/23/2026 | 698,941 | 3,547,783 | 188,967 | 3,736,750 | 19.6% |
| 10/24/2026 | 518,127 | 4,065,910 | 199,621 | 4,265,531 | 22.4% |
| 10/25/2026 | 268,428 | 4,334,338 | 202,563 | 4,536,901 | 23.8% |
| 10/26/2026 | 583,115 | 4,917,453 | 218,852 | 5,136,305 | 27.0% |
| 10/27/2026 | 551,193 | 5,468,646 | 228,936 | 5,697,582 | 29.9% |
| 10/28/2026 | 549,424 | 6,018,070 | 240,655 | 6,258,725 | 32.9% |
| 10/29/2026 | 534,130 | 6,552,200 | 251,400 | 6,803,600 | 35.7% |
| 10/30/2026 | 12,786 | 6,565,015 | 253,818 | 6,818,833 | 35.8% |
Dates before Oct 19 reflect mail ballots only (in-person early voting began Oct 19).
Key counties
| County | Registered | Cum. in person | Cum. mail | Cum. total | % reg. | 2022 full EV period | 2026 so far vs 2022 full |
|---|---|---|---|---|---|---|---|
| Harris | 2,746,177 | 861,077 | 39,020 | 900,097 | 32.8% | 752,491 | 119.6% |
| Dallas | 1,493,955 | 534,069 | 16,154 | 550,223 | 36.8% | 412,865 | 133.3% |
| Tarrant | 1,228,367 | 506,242 | 18,297 | 524,539 | 42.7% | 411,576 | 127.4% |
| Bexar | 1,325,780 | 450,256 | 23,469 | 473,725 | 35.7% | 390,042 | 121.5% |
| Travis | 945,505 | 389,665 | 15,691 | 405,356 | 42.9% | 330,658 | 122.6% |
| Collin | 788,729 | 331,642 | 7,872 | 339,514 | 43.0% | 265,712 | 127.8% |
| Denton | 688,032 | 281,794 | 6,731 | 288,525 | 41.9% | 229,224 | 125.9% |
| Fort Bend | 581,636 | 236,760 | 5,519 | 242,279 | 41.7% | 187,191 | 129.4% |
| Williamson | 472,191 | 200,833 | 7,223 | 208,056 | 44.1% | 164,666 | 126.4% |
| Montgomery | 487,323 | 161,658 | 8,266 | 169,924 | 34.9% | 140,807 | 120.7% |
| El Paso | 530,978 | 137,235 | 7,180 | 144,415 | 27.2% | 104,135 | 138.7% |
| Hidalgo | 466,208 | 144,253 | 3,986 | 148,239 | 31.8% | 102,394 | 144.8% |
2022 column = final early-vote total for the whole 2022 November early-voting period (in person + mail) from each county's SOS historical page. Same-day 2022 and 2024 comparisons are not available: the SOS's current turnout system does not serve past general elections, and the SOS county pages do not yet list 2024. A county showing very low in-person numbers (e.g., only a handful) usually means its roster has not been uploaded to SOS yet, not that turnout is low.
County grouping trend
| Report date | Dem-leaning (12) votes | % reg. | GOP-leaning (242) votes | % reg. | Dem-leaning share of votes |
|---|---|---|---|---|---|
| 10/14/2026 | 42,172 | 0.5% | 37,345 | 0.3% | 53.0% |
| 10/15/2026 | 47,007 | 0.6% | 46,587 | 0.4% | 50.2% |
| 10/16/2026 | 53,501 | 0.7% | 52,301 | 0.5% | 50.6% |
| 10/17/2026 | 54,385 | 0.7% | 55,069 | 0.5% | 49.7% |
| 10/18/2026 | 54,386 | 0.7% | 55,836 | 0.5% | 49.3% |
| 10/19/2026 | 386,836 | 4.9% | 524,214 | 4.7% | 42.5% |
| 10/20/2026 | 696,657 | 8.9% | 973,710 | 8.7% | 41.7% |
| 10/21/2026 | 986,906 | 12.6% | 1,384,999 | 12.4% | 41.6% |
| 10/22/2026 | 1,257,139 | 16.0% | 1,763,140 | 15.8% | 41.6% |
| 10/23/2026 | 1,557,674 | 19.8% | 2,179,076 | 19.5% | 41.7% |
| 10/24/2026 | 1,762,674 | 22.4% | 2,502,857 | 22.4% | 41.3% |
| 10/25/2026 | 1,878,435 | 23.9% | 2,658,466 | 23.8% | 41.4% |
| 10/26/2026 | 2,125,290 | 27.1% | 3,011,015 | 26.9% | 41.4% |
| 10/27/2026 | 2,360,261 | 30.1% | 3,337,321 | 29.8% | 41.4% |
| 10/28/2026 | 2,599,439 | 33.1% | 3,659,286 | 32.7% | 41.5% |
| 10/29/2026 | 2,802,845 | 35.7% | 4,000,755 | 35.8% | 41.2% |
| 10/30/2026 | 2,810,732 | 35.8% | 4,008,101 | 35.8% | 41.2% |
Grouping method: a county is "Dem-leaning" if Harris received more votes than Trump there in the 2024 presidential race (Bexar, Brooks, Dallas, Dimmit, El Paso, Fort Bend, Harris, Hays, Jim Hogg, Presidio, Travis, Zavala); all other 242 counties are "GOP-leaning". 2024 county results from the tonmcg county-level results dataset (secondary source). This shows where ballots are being cast, not how anyone voted. Subject to the county upload lag noted above.
Texas voter registration change
| Election | Registered voters | Change to 2026 |
|---|---|---|
| Nov 2022 | 17,672,143 | 1,369,541 (7.7%) |
| Nov 2024 | 18,623,931 | 417,753 (2.2%) |
| Nov 2026 (current) | 19,041,684 |
Texas does not register voters by party. 2022/2024: SOS registration & turnout figures; 2026: sum of county registered voters in the SOS early-voting feed.
Nightly read commentary · not data
Interpretation written by the tracker's author, separate from the official figures above. Last updated 10/29/2026, 8:10:11 PM CT.
R vs D index: method, caveats & sources
R vs D Performance Index model estimate · not a forecast
Daily formula (v3.2, from Oct 23): Index = Texas Senate model margin ÷ 2, rounded to 0.1, capped ±10 (R positive). That margin already blends early-vote data (weight 0.60 + 0.35 × share of expected votes banked, max 0.95; 0.74 tonight) with FPS-graded, bias-corrected polls + prior + ratings (the rest, 0.28 tonight), each counted once. No betting or prediction-market data. Bands follow the gauge colors: under 1 Toss-up, 1–3 Lean, 3–6 Likely, 6+ Strong. The Texas tab’s “turnout composition index” is the older primary-history measure and is shown for comparison only.
This index shows who is turning out compared with past elections, not how anyone voted. Texas doesn't register voters by party, so voters are grouped by the party primaries they've voted in before.
| Day 3 early voters by primary history (cumulative) | 2026 | 2024 same day | 2022 same day |
|---|---|---|---|
| Republican primary history (incl. mixed, R most recent) | 831,302 (36.4%) | 1,172,279 (41.2%) | 777,779 (47.7%) |
| Democratic primary history (incl. mixed, D most recent) | 920,570 (40.3%) | 753,463 (26.5%) | 529,900 (32.5%) |
| Primary-history lead | D +3.9 pts (Day 1: D +9.6) | R +14.7 pts | R +15.2 pts |
| Final statewide result that year | — | R+13.7 (President) | R+10.9 (Governor) |
Drivers (through Day 3, Oct 21). (1) The Day 3 electorate is D+3.9 by primary history (920,570 D vs 831,302 R), about 19 points more Democratic than at the same point in 2022 (R+15.2) or 2024 (R+14.7). That is why the index is near zero. (2) The gap is narrowing: it was D+9.6 on Day 1, so Republican primary voters are catching up as early voting goes on. (3) Turnout is 140% of 2022's Day 3 pace (2,284,737 votes, 12.1% of registered). (4) Hispanic-surname voters are 18% of votes (15% in 2022). Ages 18–29 are 8% of votes, 65+ are 45%, so it is still an older electorate. In 2018, heavy early turnout in the big counties still ended R+2.6. Source: Ryan Day 3 report.
Earlier composition method (until Oct 22; still shown on the Texas tab for comparison): Implied margin = baseline final margin + 0.5 × (change in primary-history lead vs the same day). The 0.5 factor is an assumption that hasn't been back-tested. It is averaged over the 2022 and 2024 baselines, and Index = margin ÷ 2. Not yet scored (gaps): county geography (state feed lags the counties), unaffiliated-voter lean, graded polls, remaining-voter headroom. Rated low confidence until 5+ days and more components are in. Primary history is a proxy: open primaries let voters cross over, and a large 2026 primary can move people between groups. Sources: Ryan Data & Research, Early Voting Report Day 3 (secondary, voter-file based); final margins: Texas SOS election results. Updated Oct 22, 2026.
Context layers not scored
A layer only joins the score if it has sourced data and beats a backtest against past results. None of these do yet.
| Layer | What the data shows | Why not scored |
|---|---|---|
| First-time / no-history voters | 6.6% of Day 10 voters (2024 Day 10: 10.0%; 2022 Day 10: 4.5%; Day 1: 3.2%; 2022 final: 7.0%; 2018 final: 12.6%) | No 2018 same-day figure; can't backtest |
| Age 18–29 share | 11.9% thru Day 10 (2024 Day 10: 11.8%; 2022 Day 10: 7.7%; Day 1: 7.2%; 2022 final: 11.5%; 2018 final: 13.3%). Day 10 groups: 18–29 11.9%, 30–44 18.2%, 45–64 34.7%, 65+ 35.2% (Ryan Day 10) | Party lean by age not modeled |
| Age 65+ share | 35.2% thru Day 10 (2024 Day 10: 31.7%; 2022 Day 10: 38.7%; Day 1: 47.9%; 2022 final: 29.0%; 2018 final: 25.7%) | Same |
| Election Day cannibalization | Estimate: about 63% or more of Day 1 voters are 2024 voters who voted later last time (90%+ voted in 2024 × ~70% of them not on Day 1) | Used only to discount the turnout surge |
| South Texas (Hidalgo, Cameron, Webb, Starr) vs 2020/2024 | Gap: same-day 2020/2024 county figures not yet on file; state feed lags | No data. Statewide Hispanic share in the Day 1 report (4.0% vs 15.3% in 2022) looks like a data-classification artifact and is not used |
| Precinct partisan mix (Dallas, Harris, Tarrant, Collin, Travis, Bexar) | Gap: precinct early-vote rosters not compiled yet | No data |
Calibration check: The 2018 electorate had a bigger Republican primary-history lead (R+13.9) than 2022 (R+10.8), yet 2018 ended R+2.6 and 2022 R+10.9. Primary-history mix alone does not reliably predict the margin (applying the 0.5 factor from 2018 predicts 2022 at R+1.1 vs actual R+10.9). So the index tracks who is showing up compared with past years, not a forecast, and confidence stays Low until a voter-lean model passes a backtest. Sources: Ryan Data & Research Day 1 report (final-election comparison pages). Method v1.1.
Tested & rejected leads (Oct 22, 2026)
- Applied: Rasmussen MN 48–45 (Oct 20–21, 1,182 LV); Fox/Beacon-Shaw ME 49–49 (Oct 16–20, 1,514 LV).
- Confirmed: Ryan Day 3 D+3.9 primary-history lead.
- Rejected: time-weighted poll averaging (backtest error 4.53 vs 4.45 pts); early-vote pace turnout effect (NC 2018/2022 backtest failed; stays off).
- Not supported: “red TX counties outpacing” (Day 3 turnout 11.2% in R-heavy vs 11.8% in D-heavy counties).
- Not verifiable with official data: Dallas Black share (no race in TX file); OH/IA Day-1 Dem collapse (daily archive started).
- Pending: WI absentee vs 2022; registration-trend method.
All states summary
All states (summary)
| State | Early votes cast | vs 2022 total turnout | Dem reg. | Rep reg. | None/other | Registers by party | Trend lean (Estimate) | As of |
|---|---|---|---|---|---|---|---|---|
| Alabama | not reported | No | — | |||||
| Alaska | 72,150 | 27.0% | — | — | — | Yes | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/29/2026 |
| Arizona | 1,541,187 | 59.5% | 575,936 | 585,570 | 379,681 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+0.6 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Arkansas | 459,961 | 50.3% | 29,618 | 40,710 | 389,633 | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| California | 5,063,243 | 45.4% | 2,422,907 | 1,350,275 | 1,007,219 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+21.2 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Colorado | 1,319,994 | 52.0% | 418,115 | 324,963 | 576,916 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+7.1 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/29/2026 |
| Connecticut | 394,385 | 30.4% | 193,882 | 75,818 | 124,685 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+29.9 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Delaware | 128,011 | 39.3% | 68,227 | 35,628 | 24,156 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+25.5 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| District of Columbia | 77,480 | 37.7% | — | — | — | Yes | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Florida | 4,686,236 | 60.1% | 1,784,052 | 1,902,063 | 1,000,121 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+2.5 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Georgia | 2,681,368 | 67.6% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Hawaii | 210,302 | 49.7% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Idaho | 234,210 | 39.3% | 21,490 | 67,894 | 22,454 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+19.8 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Illinois | 1,382,486 | 33.4% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Indiana | 620,121 | 33.0% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Iowa | 407,374 | 33.1% | 191,681 | 137,833 | 77,860 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+13.2 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Kansas | 375,937 | 37.3% | 158,666 | 161,624 | 55,647 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+0.8 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Kentucky | 192,709 | 12.8% | 95,645 | 83,726 | 13,338 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+6.2 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Louisiana | 527,104 | 37.4% | 224,143 | 207,261 | 95,700 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+3.2 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Maine | 288,672 | 42.4% | 135,647 | 66,001 | 87,024 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+24.1 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Maryland | 962,046 | 47.4% | 613,885 | 203,391 | 144,770 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+42.7 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Massachusetts | 1,070,364 | 42.6% | — | — | — | Yes | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Michigan | 1,858,731 | 41.3% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Minnesota | 728,961 | 28.9% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Mississippi | 56,049 | 7.9% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/26/2026 |
| Missouri | 260,656 | 11.3% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/29/2026 |
| Montana | 274,985 | 58.7% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Nebraska | 215,424 | 31.6% | 83,442 | 95,087 | 36,895 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+5.4 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Nevada | 665,060 | 65.0% | 263,156 | 227,239 | 174,665 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+5.4 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| New Hampshire | not reported | Yes | — | |||||
| New Jersey | 855,440 | 32.3% | 509,983 | 195,033 | 150,424 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+36.8 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| New Mexico | 378,049 | 52.9% | 200,045 | 114,441 | 63,563 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+22.6 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| New York | 1,423,100 | 23.9% | — | — | — | Yes | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| North Carolina | 2,353,234 | 62.1% | 925,125 | 603,630 | 824,479 | Yes | Leans D +7.1 vs ’22Estimate. Early-vote D-minus-R share now D+13.7 (UF Election Lab, 10/30/2026) vs D+6.5 in 2022 at the same point (4 days before Election Day; NCSBE official absentee file). Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. No 2024 same-day file is on hand. | 10/30/2026 |
| North Dakota | 74,769 | 30.8% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/29/2026 |
| Ohio | 1,485,900 | 35.4% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Oklahoma | 138,490 | 12.0% | 55,893 | 65,328 | 17,269 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+6.8 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Oregon | 1,086,787 | 54.4% | 457,137 | 337,967 | 291,683 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+11.0 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Pennsylvania | 1,008,945 | 18.6% | 654,457 | 254,870 | 99,618 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+39.6 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Rhode Island | 116,845 | 32.3% | 50,595 | 11,304 | 54,946 | Yes | No baselineEstimate. Early-vote D-minus-R share now D+33.6 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| South Carolina | 876,288 | 51.0% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| South Dakota | 66,534 | 18.8% | 19,847 | 35,557 | 11,130 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+23.6 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/29/2026 |
| Tennessee | 1,280,642 | 72.9% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Texas | 6,871,574 | 84.3% | — | — | — | No | Proxy D+0.9 · ↑R since Day 1Estimate. Texas has no party registration. Proxy: D-minus-R primary-history share of early voters from Ryan Data & Research: D+9.7 on Day 1, D+3.9 Day 3, D+2.1 Day 5, D+2.6 Day 7 (weekend uptick), D+2.0 Day 8, D+1.2 Day 9, D+0.9 Day 10. Same day in 2022: R+12.3; 2024: R+11.9 (Ryan Day 10). | 10/30/2026 |
| Utah | 408,891 | 37.7% | 85,240 | 213,216 | 110,435 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+31.3 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
| Vermont | 139,565 | 47.8% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Virginia | 1,408,954 | 46.6% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Washington | 1,513,547 | 49.3% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/29/2026 |
| West Virginia | 114,391 | 23.1% | 44,638 | 46,632 | 23,121 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+1.7 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/29/2026 |
| Wisconsin | 693,334 | 25.9% | — | — | — | No | No party dataThis state does not report early voters by party, and no real proxy is on file. No lean is shown. | 10/30/2026 |
| Wyoming | 57,968 | 29.2% | 9,917 | 43,694 | 4,357 | Yes | No baselineEstimate. Early-vote D-minus-R share now R+58.3 (UF Election Lab). No 2022 or 2024 same-day baseline is on file for this state, so no lean is claimed. Day-over-day direction starts once the site has stored a prior day for this state. Caution: Democrats front-load mail ballots, so early D leads usually shrink later (NC 2022: D+8.5 at this point, D+7.1 at the end of early voting). Republicans have voted early more since 2020, so a smaller D lead than 2022 does not by itself mean R gains. | 10/30/2026 |
Secondary source (aggregator): University of Florida Election Lab, 2026 early vote (CSV), which compiles state/county election-office reports; each state page links the original source. "Dem/Rep/None" = early voters by party registration, shown only where the state reports it. "vs 2022" compares early votes so far with the state's entire 2022 general-election turnout. Tap a column header to sort, a state for details. "Registers by party" per NCSL.
Trend lean (Estimate): the D-minus-R share of early votes now vs the same point in 2022, from official files, plus the day-to-day direction. It is shown only where a real same-day baseline exists; otherwise it says No baseline or No party data. Tap a chip for details.