Mail rejection · 2022
3.4%
Texas · 12,575 rejected
U.S. 1.5% · 549,824
Live board · snapshot September 30, 2026 · election November 3. Senate today 53R–47D · Democrats need 51 · Vance breaks 50–50. RCP generic ballot D+8.3 · Sep 30. Wall Street Journal generic D 50 – R 42 · Sep 30. Silver Bulletin likely-voter generic D+9.6 · Sep 29. Sep 30: MI El-Sayed +7 Marist · IA Hinson +1 · ME Collins +2 · OH Brown +3. Cook Senate: 7 toss-ups · AK IA ME MI NH OH TX. Charlie Cook: House pickup ~20 a done deal · Senate +4/+5 very realistic. DDHQ: House 72% D · Senate 54% D. 2022 mail rejection 1.5% nationwide · EAC EAVS · not a 2026 count. Official feeds · EAC EAVS interactive · CISA 2026 plan · USAGov election offices · no 2026 canvass published. DOJ voter-roll suits 0 for 26 decided · ProPublica Oct 16. No poll-hour orders filed · incident log opens November 3. ERIC 2024 · 701,420 cross-state movers · 203,195 in-state duplicates · 28,905 deceased matches · not dual ballots
Senate48D–52RRepublican majority7 open
1,264watching
—
Brief 9:00 AM ET
United States · 120th Congress · September 30, 2026
November 3. Republicans hold the White House and thin majorities in both chambers. Democrats need a net gain to take either one. Cook calls the Senate a true coin flip. The House is the chamber more ratings say can change hands. Call the toss-ups — the map, the chamber, and every route to 51 move together. Published ratings. Not a house model.
Countdown · November 3, 2026 · Eastern
Live results
Polls open November 3. The tape below rotates the latest published surveys. It is not a vote count, and it does not invent a new sample.
—
MI · Senate survey · Sep 30
El-Sayed51
vsRogers44
Marist. Democrat +7. Survey share, not votes cast.
Next line in 5s
Mail rejection · 2022
3.4%
Texas · 12,575 rejected
U.S. 1.5% · 549,824
Seven Cook toss-ups. Click a state to cycle it: open, Democratic, Republican. The map, the bar, and the paths use the same calls.
5 pts Republican05 pts Democratic
Adds the same number of points to each toss-up’s latest lead the next time you move it. Alaska +2.3, Iowa −1, Maine −2, Michigan +7, New Hampshire +6.4, Ohio +3, Texas +2.9. A lead that crosses zero gets called. Cook’s rating does not change.
Lead vs 2022 rejection rate
Disqualification band
0.0
218R · 214D
Plus 1 independent and 2 vacancies. Democrats need a net of 4 to reach 218. Charlie Cook, on ABC Sep 27, still called a pickup around 20 a done deal. Forbes, citing Cook on Sep 28, counted 22 House toss-ups, 15 of them Republican-held. This board does not call districts.
Decision Desk HQ, via The Hill: 72% House, 54% Senate.
22 House toss-ups · Forbes, citing Cook, Sep 28
15 Republican-held7 other
2018, 2014, and 2010 are national House vote margins after the election, not pre-election generic ballots. 2026 is still a poll average.
2016
1.0%
2018
1.4%
2020
0.8%
2022
1.5%
2026
—
Combinations, not a model
46–53 D
29 of 128 toss-up maps reach a Democratic majority. 35 land 50–50 and Vance decides them. 64 stay Republican. Uncalled leans and likelies follow Cook unless you overrode them.
The white mark is seat 51. Stripes are still uncalled — blue if they break Democratic, red if they break Republican. If toss-ups stick with the incumbent, the board reads 48D–52R.
Senate now
53R · 47D
45 Democrats plus 2 independents. Thirty-five seats are up November 3. Seven are Cook toss-ups. Democrats need a net of four. Vance breaks 50–50.
House now
218R · 214D
1 independent, 2 vacancies. Cook treats a Democratic pickup around 20 as the done deal. This board does not invent districts.
White line · 218
Generic ballot
D+8.3
RCP average, Sep 30. The bar is the Wall Street Journal the same day, 50 to 42. Silver Bulletin’s Sep 29 likely-voter generic was D+9.6.
Cook Senate range
D+1 to +5
Democrats need +4. 35 seats are rated on this cartogram. Three ratings moved toward Democrats on Sep 23.
Texas SAVE flags
2,500+
Texas flagged more than 2,500 registrations as possible noncitizens after a SAVE check. Reporting found that some of those people were citizens. This is not a count of confirmed illegal votes.
Type a postal code. A split tile is a toss-up. States in today’s brief pulse until you select one.
Drag, scroll, and double-click. A toss-up call here uses the same board as the arena.
Drag to move. Scroll to zoom. Double-click a state to center it.
Shift in the margin
↗ More Democratic↘ More Republican
Compared with the 2020 presidential result. A longer arrow means the margin moved farther.
Texas. R +8.1 since 2020
East to west, not a poll-close schedule. Mark a winner. The chamber and the path count move with you. 0 calls on the board.
No race called. Ratings only. Seven toss-ups still sit in the middle, and North Carolina is the one Lean D.
A horizon line with one mark per toss-up, lifted by the size of the lead. Left to right is the desk order of the seven, not geography. Higher is a larger September 30 lead. The gold-ringed mark is the state in focus.
Governors · Cook Oct 13
Cook’s sheet lists 18 Democratic-held seats and 18 Republican-held seats. The name is the incumbent, or “Open.” A color is who holds it now, not a vote. The only rating changes printed here are the four the Washington Post reported that day. No ballot share is filled in, because no precinct has reported.
CAD
Open
COD
Open
CTD
Lamont
HID
Green
ILD
Pritzker
MAD
Healey
MDD
Moore
MED
Open
MND
Open
NMD
Open
NYD
Hochul
PAD
Shapiro
RID
Open
AZD
Hobbs
MID
Open
ORD
Kotek
WID
Open
KSD
Open
AKR
Open
Lean D
IAR
Open
GAR
Open
NVR
Lombardo
OHR
Open
FLR
Open
TXR
Abbott
NER
Pillen
Likely R
NHR
Ayotte
OKR
Open
Likely R
VTR
Scott
Likely R
ALR
Open
ARR
Sanders
IDR
Little
SCR
Open
SDR
Rhoden
TNR
Open
WYR
Open
First-digit law · not an audit
The outline is Benford’s expected share for the first digit of a number, from 1 through 9. A real check would lay observed precinct totals on top of it. Those totals are not published, so nothing here is marked normal or suspicious.
1
30.1
2
17.6
3
12.5
4
9.7
5
7.9
6
6.7
7
5.8
8
5.1
9
4.6
This counts maps. It is not a probability, a poll model, or Cook’s forecast. Each open toss-up is a coin with two faces. Leans and likelies are already stamped to Cook, unless you called them yourself. Still open: AK, IA, ME, MI, NH, OH, TX.
Blue bars are a Democratic majority. The split bar is 50–50. Red is a Republican Senate.
Shortest paths first. Choosing one fills only the toss-ups that were still open.
Published surveys only. The tape rotates through them. It does not invent a new sample.
50–42
Wall Street Journal · Sep 30 · D+8
MI · Marist · El-Sayed 51, Rogers 44
D+7
RCP’s average on Sep 30 is D+8.3. The rows are individual surveys that published both shares.
Democrats have to turn a national wave into four Senate wins that are not gifts — hold the Democratic toss-ups and take seats the Republican ticket carried in 2024, or an equivalent mix. That is why the environment can be blue and the chamber a toss-up at the same time.
Sep 23 moves: Georgia Lean D → Likely D · Kansas Likely R → Lean R · South Carolina Solid R → Likely R · New Hampshire Lean D → Toss-Up (Sep 15)
Roll litigation
Common Cause v. DOJD.D.C.
Blocked Oct 9
Democracy DocketUnited States v. Benson6th Cir.
State won
District docketDOJ unredacted-roll suits30 states + DC
0 for 26
ProPublica, Oct 16What each side is fighting for
No candidates on this chart. Each dot is a policy one party is pushing. Left to right is the economy. Bottom to top is how much the state directs people. The percent is how many adults back that policy in a published survey.
Where the dot sits is this board’s placement, not a DW-NOMINATE score and not the poll. A Republican fight can sit on the regulated side. Higher tariffs do.
Democrats are fighting for this
86%
Cap drug prices to other countries
Favor a federal maximum taken from what the same drugs cost in other rich countries. Republicans in this sample also landed at 90%.
Program for Public Consultation · Feb.–Mar. 2026 · online opt-inDemocrats are fighting for this
83%
Child care at no more than 7% of income
Favor federal money so low- and middle-income families pay no more than 7% of income on child care.
Program for Public Consultation · Feb.–Mar. 2026 · online opt-inDemocrats are fighting for this
77%
Tax credits for zero-carbon energy
Favor tax credits for energy that emits no carbon. Republicans in this sample also landed at 81%.
Program for Public Consultation · Feb.–Mar. 2026 · online opt-inRepublicans are fighting for this
62%
A large military presence on the border
Favor keeping a large military presence at the U.S.–Mexico border.
Pew Research Center · Jan. 20–26, 2026 · 8,512 adultsDemocrats are fighting for this
60%
Abortion legal in all or most cases
Say abortion should be legal in all or most cases. On the same question, 38% say illegal in all or most cases.
Pew Research Center · Jan. 20–26, 2026Republicans are fighting for this
37%
The higher tariffs
Approve the administration’s tariff increases. 60% disapprove.
Pew Research Center · Jan. 20–26, 2026 · 8,512 adultsRepublicans are fighting for this
34%
Pause every asylum application
Favor suspending all asylum applications. 66% oppose.
Pew Research Center · Jan. 20–26, 2026 · 8,512 adultsDemocratic frame beside the Republican frame
These are the arguments each side is making this week. They are not fabricated quotations from a named article. The outlet line is where that argument usually runs.
September 30, 2026
Democratic desk
Usually MSNBC, Washington Post
Asylum rules are in court, and shelters are past capacity
The Democratic argument is about the right to ask for asylum, conditions on the Mexican side, and more legal visas.
Republican desk
Usually Fox News, National Review
Enforcement is still a loophole, not a closed border
The Republican argument is about continued entries, parole, city budgets, Remain in Mexico, and wall money.
September 29, 2026
Democratic desk
Usually New York Times, Vox
Cuts help borrowers; margins are keeping prices up
The Democratic frame blames post-pandemic shocks and corporate margins, and wants the labor market protected.
Republican desk
Usually Wall Street Journal, Breitbart
Federal spending is the hangover that can relight inflation
The Republican frame blames deficits, energy subsidies, and regulation for weaker household buying power.
September 28, 2026
Democratic desk
Usually The Atlantic, Politico
Subsidies are building factories outside the big coasts
The Democratic frame is climate targets, battery plants, and the risk of staying on oil.
Republican desk
Usually Daily Wire, Washington Examiner
Mandates cut choice and lean on a thin grid
The Republican frame is Chinese minerals, older auto plants, and blackouts at peak demand.
September 27, 2026
Democratic desk
Usually MSNBC, The Atlantic
A federal statute is the turnout argument
The Democratic desk says a federal law is how access survives state bans, and why irregular voters come out.
Republican desk
Usually Fox News, National Review
The states already decide
The Republican desk says the Court sent the question back to the states, and a federal statute would override them.
September 26, 2026
Democratic desk
Usually The New York Times, The Washington Post
Removals run into asylum and shelter math
The Democratic argument is due process, conditions on the Mexican side, and cities that cannot carry the shelter bill.
Republican desk
Usually Fox News, Daily Wire
The removal program is the point of the election
The Republican argument is a large domestic program, an end to broad parole, and no new federal shelter money.
September 25, 2026
Democratic desk
Usually The Atlantic, Politico
Alliances are cheaper than the alternative
The Democratic frame is holding NATO and the Indo-Pacific, and treating aid to Ukraine as deterrence.
Republican desk
Usually Fox News, Washington Examiner
Allies should pay, and Paris can go
The Republican frame is higher European spending, leaving the Paris accord, and less open-ended aid.
September 24, 2026
Democratic desk
Usually The New York Times, Vox
Caps and Medicare negotiation stay
The Democratic frame is a capped price and keeping Medicare’s authority to negotiate.
Republican desk
Usually The Wall Street Journal, National Review
Caps become shortages
The Republican frame says price controls thin the pipeline, and that supply matters more than a statute.
September 23, 2026
Democratic desk
Usually The Washington Post, Vox
The child credit is the middle-class cut
The Democratic frame expands the child tax credit and says the 2017 law skewed the gains upward.
Republican desk
Usually The Wall Street Journal, Fox News
Letting the cuts lapse is a tax increase
The Republican frame says households pay if those cuts expire, and a larger credit is more spending.
Official feeds
2024 EAVS comprehensive report, published June 30, 2025. 2024 datasets, version 2.0, released February 12, 2026. The 2026 survey’s public comment closed May 25, 2026. Results are not published. Open a portal for the file itself. This board does not copy a number out of a tool it cannot read.
OpenFEC · filed through Sep 30
Principal-committee receipts and cash on hand from the Federal Election Commission. The coverage date on these filings is September 30, 2026. The bar is money raised, not votes. Primary losers stay in the file; the two names are the highest Democratic and Republican receipts in that state.
AK
IA
ME
MI
NH
OH
TX
Integrity · published record only
Statutes, the 2022 Election Administration and Voting Survey, and court rulings that were already public on October 18. No house model of fraud. Empty rows are waiting on an official feed.
Explore a metric
1.5%
of returned mail ballots · 2022
549,824 ballots were rejected out of 36,683,450 returned. The path across cycles is 2016 1.0%, 2018 1.4%, 2020 0.8%, 2022 1.5%.
This is not a 2026 cure list. Alaska and New Hampshire are missing from the state extract on this board.
EAC EAVS, via Ballotpedia, Aug 14, 2023Six desks · Sep 30 file · AK
Roll counsel
Voter-roll lawsuits
AK D 2.3
Democracy DocketCanvass clerk
Recount statutes
2.3 pts
NCSLList desk
ERIC reports
2022 2434k
ERIC statisticsDigit desk
Second-digit Benford
12.0%
EAC studiesReturns desk
Counted ballots
Blocked Oct 9
Cook Senate ratingsChain desk
Modules 31–40
2.3 pts
2022 EAVS briefMail rejection tape · 2022 EAVS
3.4%
Texas rejected 12,575 mail ballots.
National rate 1.5% (549,824 of 36,683,450). Alaska and New Hampshire were not in the state extract.
Not a live cure list, and not a vote margin. The bar is the published poll lead. The mark is that state’s 2022 mail-rejection rate, in points. Crossing it means the lead is inside the historical rejection rate — not that those ballots would change the winner.
Other reason given32.7%
Signature did not match26.9%
Missed deadline23.1%
Signature’s share of rejections
BallotpediaCommon Cause v. DOJD.D.C.
Blocked Oct 9 · 1:26-cv-01352
Filed April 21, 2026. Judge Sooknanan’s Oct 9 opinion vacated the DOJ voter-list program. DOJ noticed an appeal and asked for a stay on Oct 16. No later order is in this snapshot.
United States v. Benson6th Cir.
State won · 1:25-cv-01148 · 26-1225
Filed Sept 25, 2025, after Michigan handed over a public list and refused the unredacted one. District opinion Feb 10, 2026. Sixth Circuit affirmed June 24. En banc denied Aug 14. CourtListener’s docket page was last touched Sept 15 and does not add a newer order here.
DOJ unredacted-roll suits30 states + DC
0 for 26 · Sep 2025 onward
ProPublica’s Oct 16 count: 26 district-court rulings, none won by DOJ, four still open, one state settled. Common Cause’s press release the same week said judges in 26 states plus the Sixth Circuit had rejected the demands. The two counts are both printed because they are not the same cut.
AK. Automatic only on a tie. Alaska Stat. § 15.15.460.
IA. No automatic trigger on the NCSL table retrieved for this build.
ME. Automatic only on a tie. A requested recount is free at or under 1% or 1,000 votes statewide (21-A MRS § 737-A).
MI. NCSL still prints 2,000 votes (MCL § 168.880a). Ballotpedia’s Oct 2026 page says the statewide trigger is 0.1% after a 2024 rewrite. Both are shown because the public summaries disagree.
NH. No automatic trigger on the NCSL table retrieved for this build.
OH. Automatic at 0.25% or less for statewide office. Ohio Rev. Code § 3515.011.
TX. Automatic only on a tie. Tex. Elec. Code § 216.001.
Narrows each published lead by 0.0 points, toward a tie. It does not give the rejected ballots to either party, and it does not change the Senate calls.
0.0 pts
Heritage case sample
1,567
Heritage calls this a sampling of proven cases, not a census of fraud. This board does not invent a state-by-state layer from that database. Dec 12, 2024 release.
SourceRoll removals, 2022
19.26 million
Voters removed from registration rolls nationwide in the 2022 EAVS. State removal rates are not redrawn here.
SourceTexas SAVE flags
2,500+
Texas flagged more than 2,500 registrations as possible noncitizens after a SAVE check. Reporting found that some of those people were citizens. This is not a count of confirmed illegal votes.
Source2016
1.0%
Presidential. Ballotpedia’s reading of EAVS.
2018
1.4%
Midterm.
2020
0.8%
Presidential.
2022
1.5%
549,824 of 36,683,450 returned mail ballots.
2026
—
Not published. The EAC survey comes after the election.
Forensic glass · 2026 file
AK, ME, MI are the only Cook toss-ups still on ERIC’s member list. Iowa, Ohio, and Texas left in 2023. New Hampshire was never on the current list. ERIC’s year table is national. It does not say how many of those records sat in Alaska, Maine, or Michigan, and it does not say anyone voted twice in 2026.
The newest public row is 2025: 13 members reporting, 78,570 cross-state movers, 10,112 in-state duplicates, 2,548 deceased matches. Membership grew again in 2026 when New York joined and Virginia rejoined. That is not a 2026 double-vote count. ERIC’s participation report, the one that can flag a ballot in two states, is not in this table.
Anomaly test not run
012.0
111.4
210.9
310.4
410.0
59.7
69.3
79.0
88.8
98.5
Gold bars are the expected share of each second digit, from log10. A 2026 test would compare precinct totals with that shape. The 95% line for a chi-square with 9 degrees of freedom is 16.919. Crossing it would mean the digits do not match the curve. It would not, by itself, mean the count was fabricated. No precinct statement of votes is published, so no state is over the line and no state is under it.
| Year | Members reporting | Cross-state movers | In-state duplicates | Deceased matches |
|---|---|---|---|---|
| 2022 | 31 + DC | 2,433,532 | 203,210 | 65,437 |
| 2023 | 28 + DC | 1,528,157 | 108,488 | 33,085 |
| 2024 | 21 + DC | 701,420 | 203,195 | 28,905 |
| 2025 | 13 | 78,570 | 10,112 | 2,548 |
A map of the seven Cook Senate toss-ups only. Every other state is left dark, because this chart is not a national forecast. Fill is the September 30 poll lead: blue where the Democrat is ahead, red where the Republican is ahead. The gold stroke is the state the six desks are reading right now. Length of a lead is not drawn here. Color only says who is ahead.
Blue · Democrat aheadRed · Republican aheadGold · desk focus
The same seven leads, read as distance from a tie. The inner ring is about 2 points, the middle ring about 4, the outer ring about 8. Michigan, at 7 points for El-Sayed, sits near the rim. Iowa, at 1 point for Hinson, stays close to the center. A gold ring marks the desk’s current state. Nothing on this dial is a counted ballot.
Seven dials, one per toss-up. The arc is the absolute poll lead on a scale that tops out at 8 points, so Michigan nearly fills its dial and Iowa barely leaves the start. Color still says which party holds the lead. Use it when you want the size of the lead without reading a table.
AK
IA
ME
MI
NH
OH
TX
Each toss-up is a node on a ring, tied back to the center. The line to the desk’s current state turns gold. Color is the party ahead. This is a locator, not a new measurement. The number beside the code is the September 30 lead.
A column chart of the absolute lead. The axis runs from a tie at the bottom to 8 points at the top. Taller means the published survey is farther from even. It does not say the lead will hold, and it does not include North Carolina or the other rated races. Those are outside the seven.
Benford’s expected share for the second digit of a number, from 0 through 9. The bars are the mathematical curve, not a test of any county’s vote. A real test needs a precinct statement of votes, which is not published yet. The gold bar is only the digit the desks are pointing at. The 95 percent line for a later chi-square test is 16.919 on 9 degrees of freedom. That line is a threshold, not a result.
Every Democratic lead in the seven is added together, and every Republican lead is added the same way. The bar is that mass, not a seat count and not a vote share. On the September 30 file the Democratic leads sum higher, because Michigan, New Hampshire, Ohio, Texas, and Alaska outweigh Iowa and Maine. Adding points is not the same thing as winning seats.
The same two sums, set as figures so the bar in the chart beside them can be read without squinting. Democratic points on the left, Republican points on the right. A desk change does not move these totals. They describe the whole file.
21.6Democratic points
3.0Republican points
Mail ballots rejected in the 2022 general election, from the Election Administration and Voting Survey, as read by Ballotpedia. The gold line is the national rate, 1.5 percent. Texas sits above it. Michigan, Maine, Ohio, and Iowa sit below it. Alaska and New Hampshire were not in the extract stored here, so they are absent rather than drawn at zero.
A semicircle from an 8-point Republican lead on the left to an 8-point Democratic lead on the right. The needle is the focused state’s September 30 lead, and the figure under it names that state. When the desks move, the needle moves. Parked at the bottom would mean a tie.
The focused race, written large: the postal code, the party ahead, and the lead in points. The thin bar under the figure is that lead on the same 8-point scale as the columns. It is the number the other instruments are pointing at.
AK
D +2.3
Seven cells. Opacity rises with the size of the lead, so a 1-point race is faint and a 7-point race is solid. Color is still the party. The gold outline is the desk’s state. An empty-looking cell is a small lead, not a missing state.
AK
IA
ME
MI
NH
OH
TX
ERIC’s published count of cross-state movers, by the year of the report. The tallest bar is 2022, when more member states reported. The short 2025 bar is fewer members reporting, not a sudden drop in people moving. These rows are list-maintenance flags. They are not people who voted twice.
A horizon line with one mark per toss-up, lifted by the size of the lead. Left to right is the desk order of the seven, not geography. Higher is a larger September 30 lead. The gold-ringed mark is the state in focus.
The seven leads as separate readings, each with the party and the points. This is the table the pictures are drawn from. September 30 is the vintage. A later canvass replaces these numbers in one file and every chart above redraws.
AK
2.3
Dem
IA
1
GOP
ME
2
GOP
MI
7
Dem
NH
6.4
Dem
OH
3
Dem
TX
2.9
Dem
A signature trace with two gold jaws. The gap between the jaws widens with the focused state’s 2022 mail-rejection rate. Where that state has no 2022 rate in the extract, the jaws use Texas, the highest rate among the states that do. The picture is a scale for a rate we have. It is not a county’s machine pass-threshold.
Sixteen timing marks. The gold ones count up with the focused state’s poll lead, two marks per point, so a 7-point lead lights most of the row and a 1-point lead lights two. The rest stay slate. This is a ruler for the lead, not a printer-calibration log.
AK · 5 of 16
A ring that fills clockwise from the top. Full would be an 8-point lead. The number in the center is the focused state. The fill is the size of its poll lead and nothing else. There is no memory-card timestamp in this snapshot, so the ring is not a courier chronometer.
Four of the seven, the ones with the most mail and the most attention on election night: Alaska, Michigan, Ohio, and Texas. Each tile is that state’s September 30 lead. The gold border is whichever of them the desks are on.
AK
D +2.3
MI
D +7
OH
D +3
TX
D +2.9
Alaska’s August 18 primary, the shares Ballotpedia recorded: Peltola 49.5, Dan S. Sullivan 41.4, the other Dan Sullivan 2.5, Heikes 1. Leslie’s 1.1 percent withdrew and Heikes advanced. These are primary shares. Ranked-choice rounds for November are not tabulated, so this is not an exhausted-ballot waterfall.
A gold barrier and a colored dot. The dot starts at the left and moves right as the focused lead grows, on the same 8-point scale. It shows how far that lead sits from zero. It is not a count of same-day registration cards returned as undeliverable.
The seven leads as a spectrum of bars, left to right in file order. Height is the absolute lead. Color is the party. Gold replaces the color on the focused state so you can see which bar the desks are describing.
The last published overseas-and-military file, the 2024 EAVS, not this election. The big figure is ballots transmitted. The bar splits the classified share: about 70.7 percent overseas civilians in blue, about 26.3 percent uniformed services in gold. The two shares do not add to 100 because some transmissions were unclassified. Returned and rejected counts sit under the figure. No 2026 arc is drawn.
1,327,324
Transmitted
Overseas 70.7%Uniformed 26.3%
806,743 returned · 30,401 rejected
A 24-hour face. The pale wedge is midnight to dawn, the hours a drop-box allegation would have to explain. The hand is not a clock time. It marks the focused state’s lead, from a tie at the top toward the right as the lead grows. No transfer sheet is in, so nothing on the wedge is an overnight harvest.
Two crossing boundaries, one blue and one red, and a gold mark whose position is the focused lead. The mark sliding right means a larger lead, not a ballot cast in the wrong district. No county has filed a split-precinct or wrong-ballot-style count on this board.
Wednesday, September 30
Next brief at 9:00 AM ET
Catch-up file. Polls RCP had posted by Wednesday night. The 9:00 AM ET rewrite starts Thursday.
RealClearPolitics said Wednesday that Democrats widened their generic-ballot lead to 8.3. A Wall Street Journal survey the same day had Democrats at 50 and Republicans at 42. Tuesday’s Quinnipiac was 51–39, and Economist/YouGov was 53–38. Silver Bulletin’s Sep 29 forecast page put the likely-voter generic at D+9.6.
Marist has Abdul El-Sayed at 51 and Mike Rogers at 44. Fox News has it 50–49. The Sep 28 RCP average was El-Sayed +3.0. Cook’s rating is still Toss-Up.
Quantus Insights on Wednesday: Ashley Hinson 47, Josh Turek 46. That is the other direction from the Sep 28 RCP average, which had Turek +0.5. One poll. Cook still rates the open seat Toss-Up.
Susan Collins 48, Troy Jackson 46. The Sep 28 RCP average had Jackson +0.9. Silver’s Sep 29 tipping-point note still had Maine at D+2.7. No public Cook move off Toss-Up.
USA Today/Suffolk on Wednesday: Sherrod Brown 47, Jon Husted 44. Marist on Tuesday was 51–43. The Sep 28 RCP average was Brown +3.8. Cook rating unchanged.
Fabrizio/Anzalone on Tuesday: Roy Cooper 53, Michael Whatley 42. InsiderAdvantage: Peggy Flanagan 46, Michele Tafoya 45. North Carolina stays Lean D. Minnesota stays Likely D.
No public Cook rating change on Wednesday. The Sep 23 shifts still stand. Charlie Cook’s Sep 27 ABC interview still frames a House pickup around 20 as a done deal, and a Senate gain of four or five as very realistic. Decision Desk HQ, via The Hill: 72% House, 54% Senate.
Who voted
No office can say how many Democrats and Republicans voted on November 3, because that election has not been held. The Census asks age, race, education, sex, income, and whether a person voted. It does not ask party. Party, by those same groups, comes from a survey on election night. The last one is the 2024 presidential election, not this midterm.
Census 2026 not published
Census 2024 published
Census 2022 published
Census · November 2024 · turnout
65.3%
154 million citizens 18 and older reported voting. 73.6% were registered (174 million).
Women66.9%
Men63.7%
Advanced degree82.5%
Bachelor’s degree77.2%
High school graduate52.5%
Voters ages 45–64 were 33.3% of the people who voted.
Election Day, in person39.6%
Early, in person30.7%
By mail29.0%
Tables released April 30, 2025. Report P20-590, August 2026. These are survey reports, not a count of ballots.
Census releaseAP VoteCast · 2024 · share of each side
120,000+ interviews, October 28 to November 5, 2024. A blue bar is the share of Harris voters in that group. A red bar is the share of Trump voters. They are not the share of the group that voted for either candidate, and they are not a 2026 party vote.
Age
18–29
17%15%
30–44
26%23%
45–64
31%34%
65+
27%28%
Education
High school or less
22%32%
Some college
29%33%
College degree
28%23%
Postsecondary degree
21%12%
Race or ethnicity
White
66%84%
Black
18%3%
Latino
11%9%
Another race
5%4%
Place
Urban
27%14%
Suburban
48%42%
Rural
26%44%