Who got 2026 right
We scored 484 Senate and governor polls from the final month of the campaign, Oct 4 to Nov 3, 2026, against the certified results in 58 races, plus every firm's final national generic ballot against the D +13.0 result. The grades below now set how much each poll counts in every On Point average.
The 2026 Gold Standard
Weighted six times a baseline pollQuinnipiac and AtlasIntel join. Cygnal moves to A minus and Quantus Insights to B minus after running well to the right of the results. The Wall Street Journal, with no scored 2026 polls, moves to A++ until it has a record to grade.
How the polls missed
Final 30 daysThe 2026 polls understated Democrats almost everywhere. Across the final month the typical poll had the margin 3.4 too Republican, and nonpartisan polls alone were 3.7 too Republican. Republican campaign and party polls were 5.1 too Republican; Democratic ones landed close to the result, 0.7 too Democratic, though with as wide a spread.
Outside reviews found the same pattern. Pew Research counted polls understating the Democratic margin by at least two points in 21 of 27 Senate and governor races, by about four points on average. G. Elliott Morris put the average miss at 3.5 points and traced most of it to turnout: likely voter screens built on past vote history were more Republican than the registered voter numbers in 25 of 32 cases and missed by more, 4.4 points against 2.6. A review of 412 final two week polls found 334 of them understated Democrats.
The firms that held up shared that lesson. NYT/Siena had the smallest typical miss of any firm with a full slate, AtlasIntel and CNN/SSRS beat the field race for race, and On Point, The Honest Poll, AtlasIntel and Big Data Poll were the four national finals closest to the D +13 result.
Partisan corrections for 2028
Points moved back before averaging| Who | Polls scored | Moved toward | Points |
|---|---|---|---|
| Republican campaign and party polls | 101 | Democrats | 2.2 |
| Democratic campaign and party polls | 59 | Republicans | 1.8 |
| Trafalgar | 13 | Democrats | 2.3 |
| InsiderAdvantage | 18 | Democrats | 1.6 |
| Quantus Insights | 8 | Democrats | 2.4 |
| GBAO | 4 | Republicans | 0.9 |
| Data for Progress | 8 | Republicans | 0.3 |
Each figure is how far that side's polls ran from the nonpartisan polls of the same races in the final month of 2026, trimmed of the most extreme tenth and shrunk toward zero when there are few polls. Measuring against the nonpartisan polls rather than the results keeps the correction about the sponsor: every kind of poll missed toward Republicans in 2026, and that shared miss is not something a single firm's correction should carry into 2028. Tables keep each poll as published.
Every firm scored on 2026
91 firms| Pollster | Grade | Was | Score | Race polls | Average miss | Lean | Generic final |
|---|---|---|---|---|---|---|---|
| The Honest Poll | Gold | Gold | −2.8 | D +13.0 | |||
| Big Data Poll | Gold | Gold | −1.9 | 2 | 3.8 | R 3.8 | D +12.3 |
| NYT/Siena | Gold | Gold | −1.7 | 21 | 3.0 | R 1.0 | D +11.0 |
| Quinnipiac | Gold | C | −1.4 | 3 | 4.2 | D 0.3 | D +12.0 |
| AtlasIntel | Gold | A++ | −1.3 | 18 | 4.1 | R 1.8 | D +14.1 |
| Emerson | Gold | Gold | −1.2 | 13 | 6.6 | R 3.2 | D +10.0 |
| On Point Polling | Gold | Gold | −1.1 | 20 | 4.5 | R 3.0 | D +13.1 |
| CNN/SSRS | A+ | C | −0.8 | 12 | 3.4 | R 2.6 | D +10.0 |
| Cook/GS Strategy | A+ | NR | −0.8 | 6 | 3.4 | R 2.7 | |
| Peak Insights R sponsor | A | NR | −2.4 | 2 | 4.4 | R 4.4 | |
| Chism Strategies D sponsor | A | NR | −1.9 | 2 | 1.3 | D 1.3 | |
| State Navigate | A | NR | −1.7 | 4 | 3.9 | R 0.3 | |
| HighGround | A | C | −1.5 | 2 | 2.3 | R 0.1 | |
| Mason-Dixon | A | D- | −1.2 | 4 | 4.8 | R 0.2 | |
| TIPP | A | A | −1.0 | D +10.0 | |||
| St. Pete Polls | A | NR | −1.0 | 4 | 1.1 | R 0.1 | |
| Beacon/Shaw R sponsor | A | C+ | −0.8 | 1 | 2.2 | D 2.2 | D +9.0 |
| Verasight | A | NR | −0.7 | 6 | 4.0 | R 4.0 | D +11.0 |
| Prolific | A | NR | −0.7 | D +10.7 | |||
| Marquette University | A | B | −0.6 | 1 | 1.4 | R 1.4 | |
| UNF | A | NR | −0.6 | 2 | 0.9 | R 0.9 | |
| Research Co. | A | C+ | −0.5 | 21 | 5.5 | R 5.0 | |
| Resonance Media | A | NR | −0.5 | 2 | 3.4 | R 3.4 | |
| YouGov | A- | C | −0.5 | 53 | 4.6 | R 3.7 | D +10.0 |
| Suffolk | A- | B+ | −0.5 | 8 | 3.7 | R 3.4 | |
| Washington Post/Schar | A- | A+ | −0.4 | 4 | 4.8 | R 0.1 | |
| Data for Progress D sponsor | A- | C- | −0.4 | 8 | 5.3 | R 4.9 | |
| MDW Communications D sponsor | A- | NR | −0.3 | 2 | 1.6 | R 1.6 | |
| Cygnal R sponsor | A- | Gold | −0.3 | 2 | 6.2 | R 6.2 | D +9.0 |
| Hart Research D sponsor | B+ | C+ | −0.3 | 2 | 2.5 | R 0.7 | |
| UGA SPIA | B+ | A+ | −0.3 | 2 | 9.8 | R 9.8 | |
| Siena College | B+ | C | −0.2 | 3 | 6.5 | R 6.5 | |
| SurveyUSA | B+ | C+ | −0.2 | 6 | 4.8 | R 4.8 | |
| Fabrizio Lee R sponsor | B+ | A | −0.2 | 1 | 4.7 | R 4.7 | |
| Impact Research D sponsor | B+ | NR | −0.2 | 2 | 4.0 | D 0.5 | |
| Mitchell Research | B+ | B+ | −0.2 | 2 | 2.9 | R 2.5 | |
| National Public Affairs R sponsor | B+ | NR | −0.2 | 12 | 4.0 | R 1.2 | |
| UC Berkeley IGS | B+ | B | −0.2 | 2 | 4.5 | R 4.5 | |
| Saint Anselm | B+ | C+ | −0.1 | 2 | 1.8 | R 0.7 | |
| Glengariff | B+ | C+ | −0.1 | 2 | 1.9 | R 1.9 | |
| Workbench Strategy D sponsor | B+ | NR | −0.1 | 3 | 2.1 | D 2.1 | |
| UMass Lowell | B+ | B- | −0.1 | D +9.0 | |||
| UT Texas Politics Project | B | NR | −0.1 | 2 | 6.4 | R 6.4 | |
| GQR D sponsor | B | NR | +0.0 | 2 | 5.9 | R 5.9 | |
| Global Strategy Group D sponsor | B | A- | +0.0 | 3 | 5.3 | R 0.5 | |
| Echelon Insights R sponsor | B- | B | +0.1 | 12 | 4.7 | R 3.6 | D +5.0 |
| Guidant Polling R sponsor | B- | NR | +0.2 | 3 | 6.0 | R 6.0 | |
| GrayHouse R sponsor | B- | NR | +0.2 | 6 | 4.8 | R 3.7 | |
| RedBox Research D sponsor | B- | NR | +0.2 | 13 | 4.2 | R 4.1 | |
| Franklin & Marshall | B- | C- | +0.2 | 1 | 1.5 | D 1.5 | |
| Quantus Insights | B- | Gold | +0.3 | 8 | 9.0 | R 8.9 | |
| Spry Strategies R sponsor | C+ | C+ | +0.3 | 1 | 9.0 | R 9.0 | |
| Texas Public Opinion Research D sponsor | C+ | NR | +0.4 | 2 | 7.9 | R 7.9 | |
| RMG Research | C+ | C+ | +0.4 | D +8.0 | |||
| Research & Polling | C+ | A | +0.4 | 2 | 8.9 | R 8.9 | |
| SurveyMonkey | C+ | D- | +0.5 | D +10.0 | |||
| CBS News | C+ | NR | +0.5 | D +8.0 | |||
| Pew | C+ | NR | +0.5 | D +8.0 | |||
| Harper Polling R sponsor | C | C+ | +0.5 | 1 | 9.6 | R 9.6 | |
| Wick Insights R sponsor | C | B- | +0.6 | 1 | 5.3 | R 5.3 | |
| Marist | C | C | +0.6 | 14 | 6.2 | R 6.2 | D +8.0 |
| East Carolina | C | A- | +0.7 | 2 | 9.5 | R 9.5 | |
| Morning Consult | C | C- | +0.7 | 2 | 5.0 | R 5.0 | D +8.0 |
| GBAO D sponsor | C | F | +0.7 | 4 | 3.7 | D 0.0 | |
| Change Research D sponsor | C | D | +0.7 | 3 | 4.6 | R 3.8 | |
| InsiderAdvantage R sponsor | C | A+ | +0.7 | 18 | 5.4 | R 5.0 | |
| UH Hobby School | C | NR | +0.8 | 2 | 8.9 | R 8.9 | |
| Noble Predictive Insights | C | C+ | +0.8 | 2 | 7.9 | R 7.9 | |
| OnMessage R sponsor | C- | A+ | +0.8 | 1 | 15.4 | R 15.4 | |
| Fox News | C- | NR | +0.9 | 6 | 7.4 | R 7.2 | D +9.0 |
| Rasmussen Reports R sponsor | C- | NR | +0.9 | 19 | 4.7 | R 4.7 | D +6.0 |
| NBC/Hart/POS | C- | NR | +0.9 | D +7.0 | |||
| The Citadel | C- | NR | +1.0 | 2 | 7.5 | R 7.5 | |
| Ipsos | C- | C | +1.0 | D +7.3 | |||
| Political Matrix | C- | F | +1.0 | 2 | 1.6 | R 1.6 | |
| John Zogby Strategies | D | NR | +1.4 | 1 | 4.6 | R 4.6 | D +6.0 |
| Leger | D | C | +1.5 | D +6.0 | |||
| Victory Insights R sponsor | D | D+ | +1.5 | 2 | 6.3 | R 6.3 | |
| Stetson | D | NR | +1.5 | 4 | 5.9 | R 5.9 | |
| Public Policy Polling D sponsor | D | D- | +1.6 | 2 | 4.5 | D 4.5 | |
| McLaughlin & Associates R sponsor | D | B | +1.6 | 1 | 12.0 | R 12.0 | |
| Alaska Survey Research | D | Super F | +1.7 | 3 | 10.3 | R 8.3 | |
| Focaldata | D | D- | +1.8 | D +7.0 | |||
| Clarity | D | Super F | +1.9 | D +8.0 | |||
| Harris Poll | D | B | +1.9 | D +4.0 | |||
| Penn State | D | NR | +2.2 | D +4.0 | |||
| UNH | D | C- | +2.3 | 4 | 6.8 | D 6.8 | |
| Trafalgar R sponsor | D- | A- | +1.9 | 13 | 8.0 | R 8.0 | |
| co/efficient R sponsor | F | C+ | +2.3 | 8 | 6.7 | R 6.7 | |
| Targoz R sponsor | F | C+ | +2.5 | 6 | 6.4 | R 6.4 | D +3.0 |
| Concord Public Opinion Partners D sponsor | F | NR | +4.8 | 10 | 13.3 | D 13.0 |
Score is a firm's miss against the other polls of the same races, in points; below zero beat the field. A national generic ballot final counts as three race polls. The old grade counts as four more polls, so a short 2026 record leans on the track record and a long one is graded on 2026. Firms with fewer than five scored polls can grade no higher than A and no lower than D, unless a Gold firm's national final landed within a point. Firms with no scored 2026 polls keep their grade.
Sources
- Certified results and On Point's poll database, final 30 days of the 2026 campaign.
- Pew Research Center, a first look at polling performance in 2026
- Strength In Numbers, the polls missed turnout
- PlusMinus4, a detailed first look at polling error in 2026
- RealClearPolitics, national polls missed a blue wave