Pollsters
Every pollster graded on its final poll before each election, over the full record 1998–2025 — not capped to the forecaster era. The grade is skill vs. the field: how its poll compared to other pollsters on the same races, so no one is rewarded for polling easy blowouts. The VotePredictor model (highlighted) aggregates all of them, so it tops the board — that's the value of averaging. Pick an office; click a column to sort, or a pollster's name for its accuracy by time to election and its record by cycle.
Difficulty-adjusted skill, with uncertainty
Our main rating. A Bayesian two-way model separates race difficulty (from how every pollster did on each race) from each pollster's skill, with a 90% credible interval (the bar) that's wide when a pollster has few final calls. Ranked by a conservative estimate, so a great average over a handful of races can't outrank a proven one — and VotePredictor, which forecasts every race, has the tightest interval. Pick an office; lower (left) miss is better.
| # | Pollster | Races | Avg miss (pts) | 90% range (wider = less data) |
|---|---|---|---|---|
| 1 | VotePredictorour model | 1920 | 5.0 | |
| 2 | Selzer & Co. | 45 | 4.8 | |
| 3 | AtlasIntel | 38 | 4.9 | |
| 4 | SurveyUSA | 523 | 5.2 | |
| 5 | Emerson College | 241 | 5.3 | |
| 6 | Siena College | 75 | 5.1 | |
| 7 | Cygnal | 56 | 5.1 | |
| 8 | The New York Times/Siena College | 147 | 5.3 | |
| 9 | Research & Polling Inc. | 45 | 5.1 | |
| 10 | Mason-Dixon Polling & Strategy | 361 | 5.5 | |
| 11 | Public Policy Polling | 300 | 5.5 | |
| 12 | Harris Insights & Analytics | 127 | 5.5 | |
| 13 | Research Co. | 81 | 5.6 | |
| 14 | EPIC-MRA | 38 | 5.5 | |
| 15 | Dan Jones & Associates | 42 | 5.5 | |
| 16 | Scott Rasmussen (Rasmussen Reports) | 370 | 5.8 | |
| 17 | University of Connecticut | 25 | 5.5 | |
| 18 | Monmouth University Polling Institute | 109 | 5.8 | |
| 19 | Rasmussen Reports | 126 | 5.8 | |
| 20 | YouGov | 433 | 5.9 | |
| 21 | Trafalgar Group | 88 | 5.8 | |
| 22 | Beacon Research/Shaw & Co. Research | 61 | 5.8 | |
| 23 | RT Strategies | 54 | 5.7 | |
| 24 | Tarrance Group | 63 | 5.8 | |
| 25 | Civiqs | 53 | 5.8 | |
| 26 | Suffolk University | 131 | 5.9 | |
| 27 | Pan Atlantic Research | 36 | 5.7 | |
| 28 | Mitchell Research & Communications | 25 | 5.7 | |
| 29 | Grove Insight | 45 | 5.8 | |
| 30 | ABC News/The Washington Post | 48 | 5.8 | |
| 31 | Vox Populi Polling | 26 | 5.7 | |
| 32 | Big Village | 92 | 5.9 | |
| 33 | We Ask America | 50 | 5.8 | |
| 34 | Marist College | 146 | 6.0 | |
| 35 | Elway Research | 26 | 5.8 | |
| 36 | Susquehanna Polling & Research Inc. | 52 | 5.9 | |
| 37 | Gallup | 38 | 5.9 | |
| 38 | Fleming & Associates | 25 | 5.8 | |
| 39 | Quinnipiac University | 152 | 6.1 | |
| 40 | Data for Progress | 70 | 6.0 | |
| 41 | Morning Consult | 50 | 6.1 | |
| 42 | Global Strategy Group | 83 | 6.1 | |
| 43 | Fairleigh Dickinson University | 31 | 6.0 | |
| 44 | Ipsos | 122 | 6.2 | |
| 45 | University of New Hampshire Survey Center | 75 | 6.2 | |
| 46 | SSRS | 51 | 6.2 | |
| 47 | Market Shares Corp. | 33 | 6.1 | |
| 48 | Change Research | 97 | 6.3 | |
| 49 | GBAO | 41 | 6.2 | |
| 50 | OnMessage Inc. | 49 | 6.3 | |
| 51 | RMG Research | 30 | 6.2 | |
| 52 | Montana State University Billings | 29 | 6.2 | |
| 53 | Clarity Campaign Labs | 27 | 6.2 | |
| 54 | Franklin & Marshall College Center for Opinion Research | 33 | 6.3 | |
| 55 | Magellan Strategies | 25 | 6.2 | |
| 56 | The Polling Company Inc. | 25 | 6.2 | |
| 57 | InsiderAdvantage | 74 | 6.4 | |
| 58 | WPA Intelligence | 33 | 6.3 | |
| 59 | Los Angeles Times | 27 | 6.3 | |
| 60 | National Research | 30 | 6.3 | |
| 61 | Remington Research Group | 40 | 6.4 | |
| 62 | PSB Research | 48 | 6.4 | |
| 63 | GQR | 86 | 6.5 | |
| 64 | Benenson Strategy Group | 34 | 6.4 | |
| 65 | Impact Research | 75 | 6.5 | |
| 66 | DCCC Targeting and Analytics Department | 35 | 6.5 | |
| 67 | St. Pete Polls | 26 | 6.4 | |
| 68 | Gravis Marketing | 125 | 6.7 | |
| 69 | Zogby Analytics | 191 | 6.7 | |
| 70 | SurveyMonkey | 156 | 6.7 | |
| 71 | Harstad Strategic Research | 25 | 6.5 | |
| 72 | Swayable | 43 | 6.7 | |
| 73 | ccAdvertising | 82 | 6.9 | |
| 74 | Garin-Hart-Yang Research Group | 86 | 6.9 | |
| 75 | Normington, Petts & Associates | 49 | 6.9 | |
| 76 | Public Opinion Strategies | 162 | 7.2 | |
| 77 | American Research Group | 203 | 7.3 | |
| 78 | Moore Information Group | 39 | 7.1 | |
| 79 | Targoz Market Research | 28 | 7.0 | |
| 80 | Digital Research | 37 | 7.1 | |
| 81 | Lucid | 26 | 7.1 | |
| 82 | Redfield & Wilton Strategies | 33 | 7.3 | |
| 83 | Lake Research Partners | 36 | 7.3 | |
| 84 | Mellman Group | 43 | 7.5 | |
| 85 | McLaughlin & Associates | 64 | 7.9 |
Full table — grades, house effects & field score
The complete record with letter grades, house effects, and skill vs. the field (the office-balanced overall). Click a column to sort, or a pollster's name to drill in.
| #▲ | Pollster | Grade | Races | vs field | Avg miss | House effect | Called right |
|---|---|---|---|---|---|---|---|
| 1 | Dan Jones & Associates | A+ | 24 | -2.2 | 5.2 | D+2.6 | 96% |
| 2 | Siena College | A | 45 | -2.2 | 5.9 | R+3.0 | 80% |
| 3 | SurveyUSA | A | 172 | -1.7 | 5.1 | R+0.5 | 83% |
| 4 | The New York Times/Siena College | A | 68 | -1.6 | 4.1 | R+1.3 | 79% |
| 5 | Cooper & Secrest | A | 12 | -1.4 | 6.3 | D+0.6 | 75% |
| 6 | RT Strategies | A- | 54 | -1.0 | 5.9 | D+1.8 | 70% |
| 7 | We Ask America | A- | 19 | -1.0 | 4.8 | R+1.9 | 84% |
| 8 | VotePredictor model | A- | 1011 | -0.9 | 6.1 | R+0.2 | 84% |
| 9 | Mason-Dixon Polling & Strategy | A- | 63 | -0.8 | 5.2 | D+0.8 | 86% |
| 10 | Public Policy Polling | B+ | 68 | -0.7 | 6.0 | D+2.3 | 77% |
| 11 | Franklin & Marshall College Center for Opinion Research | B+ | 13 | -0.7 | 5.4 | R+0.5 | 85% |
| 12 | American Viewpoint | B+ | 15 | -0.6 | 7.2 | R+5.6 | 67% |
| 13 | Hendrix College | B+ | 14 | -0.5 | 8.0 | D+2.7 | 93% |
| 14 | Research & Polling Inc. | B+ | 24 | -0.4 | 5.1 | D+0.1 | 88% |
| 15 | The Polling Company Inc. | B+ | 16 | -0.3 | 5.6 | R+3.3 | 69% |
| 16 | Tarrance Group | B+ | 42 | -0.3 | 6.3 | R+2.7 | 69% |
| 17 | University of New Hampshire Survey Center | B | 32 | -0.3 | 6.4 | D+1.5 | 78% |
| 18 | GBAO | B | 27 | -0.3 | 5.7 | D+3.5 | 67% |
| 19 | Pan Atlantic Research | B | 15 | -0.2 | 5.9 | D+4.6 | 87% |
| 20 | Zogby Analytics | B | 40 | -0.2 | 7.4 | R+1.1 | 70% |
| 21 | Grove Insight | B | 29 | -0.1 | 5.9 | D+4.3 | 86% |
| 22 | Normington, Petts & Associates | B | 40 | -0.1 | 8.3 | D+6.6 | 55% |
| 23 | Global Strategy Group | B | 49 | +0.0 | 6.4 | D+4.0 | 71% |
| 24 | Monmouth University Polling Institute | B | 34 | +0.0 | 6.3 | D+0.8 | 79% |
| 25 | PSB Research | B- | 45 | +0.0 | 6.9 | D+5.2 | 76% |
| 26 | Emerson College | B- | 50 | +0.1 | 5.6 | R+1.3 | 84% |
| 27 | Change Research | B- | 40 | +0.2 | 6.2 | D+4.3 | 73% |
| 28 | St. Pete Polls | B- | 20 | +0.2 | 7.3 | D+1.0 | 80% |
| 29 | Benenson Strategy Group | B- | 25 | +0.3 | 7.1 | D+4.2 | 60% |
| 30 | National Research | B- | 13 | +0.3 | 6.7 | R+5.2 | 62% |
| 31 | Gravis Marketing | B- | 15 | +0.3 | 10.4 | R+4.9 | 73% |
| 32 | Impact Research | C+ | 56 | +0.4 | 6.8 | D+5.8 | 68% |
| 33 | Susquehanna Polling & Research Inc. | C+ | 20 | +0.5 | 8.0 | D+3.7 | 75% |
| 34 | ccAdvertising | C+ | 54 | +0.5 | 7.7 | R+5.9 | 82% |
| 35 | WPA Intelligence | C+ | 17 | +0.6 | 6.5 | R+4.5 | 77% |
| 36 | OnMessage Inc. | C+ | 18 | +0.6 | 7.1 | R+4.4 | 72% |
| 37 | DCCC Targeting and Analytics Department | C+ | 33 | +0.6 | 6.3 | D+5.0 | 61% |
| 38 | Garin-Hart-Yang Research Group | C | 26 | +0.7 | 7.7 | D+6.0 | 69% |
| 39 | Moore Information Group | C | 14 | +0.8 | 9.0 | R+3.4 | 64% |
| 40 | GQR | C | 44 | +0.8 | 7.1 | D+5.0 | 73% |
| 41 | MRG Research | C | 12 | +0.9 | 7.8 | R+4.7 | 75% |
| 42 | Digital Research | C- | 15 | +1.0 | 8.2 | D+3.0 | 87% |
| 43 | Public Opinion Strategies | C- | 96 | +1.2 | 7.9 | R+4.4 | 66% |
| 44 | Lake Research Partners | C- | 21 | +1.2 | 8.3 | D+6.6 | 71% |
| 45 | Mellman Group | D | 16 | +1.3 | 8.7 | D+6.6 | 63% |
| 46 | McLaughlin & Associates | D | 26 | +2.9 | 11.9 | R+9.7 | 50% |
- Grade — letter rating, higher is better (A+ best), from the skill-vs-field score.
- vs field — points better/worse than other pollsters on the same races (leave-one-out). Negative is better — it beat the field. This is what the grade ranks on.
- Avg miss — each pollster's actual average miss in points across its final calls, for context. Lower is better, but it's heavily influenced by which races they polled (easy vs. close) and by sample size — the Grade / vs-field columns adjust for that, this one doesn't.
- Races — races scored: a pollster's final poll in each (the model's forecast in each). President is counted by state.
- House effect — average partisan lean; D+ overstates Democrats, R+ overstates Republicans.
- Called right — share of races where the winner was called. Intuitive, but mostly reflects how lopsided the races were, not skill.
The VotePredictor row (highlighted) is our own forecast, graded the same way. Its vs-field score is its prediction vs. the pollsters who polled those same races — it beats the average pollster (it aggregates them, so that's expected). Its absolute Avg error looks higher because it forecasts every race, including the House-heavy, hardest-to-call set — which is exactly why grading on skill-vs-field, not raw error, is the fairer comparison. To rank pollsters against forecasters head-to-head, see the combined board.
Compare on common races
The fairest test when coverage differs: pick pollsters and they're all scored on the exact same races — the ones they all polled (the head-to-head, generalized), by final-poll margin error. VotePredictor forecasts every race, so adding it never shrinks the set; adding a pollster that polled few races does. The list shows the most prolific pollsters.
Everyone scored on the 99 races all 4 selected cover — identical races, exactly apples-to-apples.
| # | Name | Avg miss (pts) ▾ | Called right |
|---|---|---|---|
| 1 | VotePredictor | 2.8 | 93% |
| 2 | SurveyUSA | 3.8 | 91% |
| 3 | YouGov | 4.1 | 93% |
| 4 | Scott Rasmussen (Rasmussen Reports) | 4.2 | 85% |