Pollster ratings
How accurate each pollster has been in the final three weeks of Senate, House, governor and presidential races since 1998, compared with other pollsters in the same race. Negative “error vs. peers” means more accurate than the field. 131 of the 196 pollsters with 2026 polls have a track record here; the rest get a slightly below-average weight until they build one.
85 pollsters
| Pollster | Grade | Polls rated | Error vs. peers | Lean | Transparency | Weight |
|---|---|---|---|---|---|---|
| Cygnal 2014–2022 | A+ | 56 | -1.65 | D+1.2 | 3.7 | 1.40× |
| AtlasIntel · AAPOR/Roper 2020–2022 | A+ | 24 | -1.38 | D+0.6 | 5 | 1.40× |
| The Washington Post/George Mason University Schar School of Policy and Government · AAPOR/Roper 2016–2021 | A+ | 11 | -1.16 | D+1.4 | 9 | 1.40× |
| American Viewpoint 1998–2022 | A+ | 23 | -1.64 | R+3.2 | 1 | 1.40× |
| Marquette University Law School · AAPOR/Roper 2012–2022 | A | 42 | -0.74 | D+1.6 | 10 | 1.33× |
| Beacon Research/Shaw & Co. Research · AAPOR/Roper 2008–2022 | A | 141 | -0.84 | D+2.0 | 5.2 | 1.28× |
| ABC News/The Washington Post · AAPOR/Roper 1998–2022 | A | 145 | -0.63 | D+1.0 | 9 | 1.28× |
| The New York Times/Siena College · AAPOR/Roper 2016–2022 | A | 192 | -0.65 | D+1.3 | 8 | 1.27× |
| Hendrix College 2010–2022 | A | 24 | -0.93 | D+6.0 | 3.9 | 1.26× |
| Marist College · AAPOR/Roper 1998–2022 | A | 330 | -0.57 | D+0.1 | 8.7 | 1.26× |
| Siena College · AAPOR/Roper 2002–2022 | A | 135 | -0.67 | R+2.1 | 6.2 | 1.25× |
| University of Massachusetts Lowell Center for Public Opinion · AAPOR/Roper 2012–2022 | A | 46 | -0.55 | D+1.5 | 8 | 1.24× |
| MassINC Polling Group · AAPOR/Roper 2010–2022 | A− | 46 | -0.62 | D+0.2 | 5.1 | 1.22× |
| DHM Research 2000–2022 | A− | 26 | -0.71 | R+1.2 | 5 | 1.21× |
| Leger 2016–2020 | A− | 27 | -0.59 | D+0.3 | 7.3 | 1.21× |
| Data for Progress 2019–2022 | A− | 176 | -0.68 | D+2.5 | 5.5 | 1.21× |
| Harper Polling 2013–2020 | A− | 27 | -0.76 | D+2.7 | 3.7 | 1.21× |
| Research & Polling Inc. 1998–2022 | A− | 77 | -0.73 | R+1.2 | 4.3 | 1.21× |
| Public Policy Polling 2008–2022 | A− | 771 | -0.72 | D+0.6 | 4 | 1.20× |
| Emerson College · AAPOR/Roper 2013–2022 | A− | 373 | -0.44 | D+0.3 | 7.5 | 1.20× |
| Echelon Insights 2016–2022 | A− | 15 | -0.57 | D+1.4 | 6.9 | 1.20× |
| Research Co. 2018–2022 | A− | 59 | -0.61 | D+1.0 | 6 | 1.20× |
| TIPP Insights 2000–2022 | A− | 47 | -0.67 | R+1.1 | 4.1 | 1.19× |
| SurveyUSA 1998–2022 | A− | 1478 | -0.50 | R+0.1 | 7.4 | 1.19× |
| RMG Research 2020–2022 | A− | 48 | -0.59 | D+4.6 | 5.5 | 1.19× |
| YouGov 2004–2022 | A− | 864 | -0.40 | D+1.2 | 8.3 | 1.17× |
| Mason-Dixon Polling & Strategy 1998–2023 | A− | 692 | -0.52 | R+1.1 | 5.6 | 1.17× |
| East Carolina University Center for Survey Research · AAPOR/Roper 2020–2022 | A− | 21 | -0.36 | D+2.9 | 6.7 | 1.16× |
| University of North Florida Public Opinion Research Lab · AAPOR/Roper 2004–2022 | A− | 22 | -0.11 | D+4.2 | 9.2 | 1.13× |
| Suffolk University · AAPOR/Roper 2002–2022 | A− | 196 | -0.15 | D+1.5 | 8 | 1.12× |
| CBS News · AAPOR/Roper 1998–2018 | B+ | 38 | -0.27 | D+1.5 | 5.3 | 1.12× |
| Susquehanna Polling & Research Inc. 2006–2022 | B+ | 79 | -0.34 | D+0.2 | 5.2 | 1.11× |
| Glengariff Group 2006–2022 | B+ | 32 | -0.52 | D+1.3 | 1.4 | 1.10× |
| Remington Research Group 2012–2022 | B+ | 76 | -0.23 | D+2.6 | 6.9 | 1.10× |
| Elon University · AAPOR/Roper 2008–2016 | B+ | 20 | -0.07 | D+1.6 | 7.6 | 1.09× |
| Hart Research Associates/Public Opinion Strategies · AAPOR/Roper 1998–2022 | B+ | 77 | -0.21 | D+2.2 | 4.4 | 1.08× |
| Trafalgar Group 2016–2023 | B+ | 167 | -0.44 | R+2.9 | 1.8 | 1.08× |
| Alaska Survey Research 2004–2022 | B+ | 36 | -0.24 | D+2.7 | 4.9 | 1.07× |
| HighGround Inc. 2016–2022 | B+ | 10 | -0.30 | R+0.3 | 3.6 | 1.07× |
| Saint Anselm College Survey Center 2008–2022 | B+ | 25 | -0.18 | D+1.8 | 6 | 1.07× |
| Quinnipiac University · AAPOR/Roper 1998–2022 | B+ | 483 | +0.13 | D+0.2 | 9.6 | 1.06× |
| University of New Hampshire Survey Center · AAPOR/Roper 2000–2022 | B+ | 213 | +0.17 | D+2.2 | 9.8 | 1.05× |
| The Winston Group · AAPOR/Roper 2004–2022 | B+ | 29 | -0.09 | D+2.4 | 4.1 | 1.05× |
| EPIC-MRA 1998–2022 | B+ | 103 | -0.22 | R+0.4 | 3.4 | 1.04× |
| Global Strategy Group 2004–2022 | B+ | 95 | -0.29 | D+3.3 | 1.7 | 1.04× |
| Tarrance Group 1998–2022 | B+ | 68 | -0.32 | R+0.8 | 1 | 1.03× |
| Ipsos · AAPOR/Roper 2002–2022 | B | 466 | +0.06 | D+3.0 | 6 | 1.03× |
| Rasmussen Reports 2012–2022 | B | 264 | -0.05 | D+1.3 | 4.6 | 1.01× |
| Clarity Campaign Labs 2012–2022 | B | 27 | -0.15 | D+5.3 | 2.3 | 1.01× |
| Fabrizio, Lee & Associates 2008–2022 | B | 15 | -0.05 | R+1.8 | 4.3 | 1.00× |
| co/efficient 2019–2022 | B | 21 | +0.02 | R+4.0 | 4.3 | 0.98× |
| National Research 2008–2021 | B | 44 | +0.03 | R+1.3 | 3.9 | 0.98× |
| Morning Consult 2016–2022 | B | 165 | +0.08 | D+4.0 | 4.8 | 0.97× |
| Tulchin Research 2010–2020 | B | 13 | -0.11 | D+1.8 | 1 | 0.97× |
| Fabrizio Lee & Associates/Impact Research 2022–2022 | B | 15 | 0.00 | R+1.2 | 2.1 | 0.96× |
| Pew Research Center · AAPOR/Roper 1998–2022 | B | 55 | +0.41 | D+1.6 | 7.9 | 0.95× |
| University of Massachusetts Department of Political Science 1998–2000 | B | 13 | +0.17 | — | 4.8 | 0.94× |
| GQR · AAPOR/Roper 1998–2022 | B− | 169 | +0.23 | D+3.3 | 2.9 | 0.93× |
| St. Pete Polls 2012–2022 | B− | 43 | +0.36 | R+0.9 | 7.1 | 0.92× |
| InsiderAdvantage 2004–2022 | B− | 253 | +0.30 | R+0.6 | 4.8 | 0.91× |
| GBAO 2010–2022 | B− | 52 | +0.14 | D+3.4 | 1.4 | 0.90× |
| Wick 2017–2022 | B− | 27 | +0.28 | R+3.7 | 4.1 | 0.90× |
| Normington, Petts & Associates 2002–2020 | B− | 49 | +0.18 | D+5.3 | 1.9 | 0.90× |
| The Tyson Group 2004–2020 | B− | 30 | +0.28 | R+0.4 | 3.7 | 0.90× |
| Victory Research 2000–2022 | B− | 17 | +0.30 | R+1.6 | 4 | 0.89× |
| Change Research · AAPOR/Roper 2017–2022 | B− | 145 | +0.42 | D+3.2 | 3.9 | 0.89× |
| High Point University Survey Research Center · AAPOR/Roper 2010–2020 | B− | 16 | +0.53 | D+3.2 | 5.8 | 0.88× |
| Franklin & Marshall College Center for Opinion Research · AAPOR/Roper 2004–2022 | B− | 61 | +0.69 | D+0.2 | 8.9 | 0.88× |
| SSRS · AAPOR/Roper 1998–2022 | B− | 74 | +0.70 | D+3.6 | 8.8 | 0.88× |
| WPA Intelligence 2002–2022 | B− | 47 | +0.43 | D+0.0 | 3.9 | 0.86× |
| Mitchell Research & Communications 1998–2022 | B− | 97 | +0.50 | D+0.2 | 5.4 | 0.85× |
| Noble Predictive Insights 2016–2022 | B− | 26 | +0.50 | D+0.1 | 5.2 | 0.85× |
| KFF · AAPOR/Roper 2016–2022 | C+ | 14 | +0.88 | D+3.6 | 9.5 | 0.83× |
| DCCC Targeting and Analytics Department 2012–2022 | C+ | 32 | +0.45 | D+4.8 | 2.5 | 0.83× |
| Impact Research 2004–2022 | C+ | 90 | +0.40 | D+4.4 | 1 | 0.82× |
| Public Opinion Strategies · AAPOR/Roper 1998–2022 | C+ | 202 | +0.54 | R+2.6 | 1.4 | 0.81× |
| Targoz Market Research 2018–2022 | C+ | 29 | +0.97 | D+2.5 | 6.5 | 0.73× |
| OnMessage Inc. 2002–2022 | C | 39 | +0.80 | R+4.0 | 2.5 | 0.72× |
| American Pulse Research & Polling 2018–2022 | C | 12 | +0.97 | D+4.2 | 3.5 | 0.69× |
| EMC Research 2000–2020 | C | 33 | +1.06 | D+5.1 | 4.4 | 0.67× |
| Lake Research Partners 2002–2022 | C | 37 | +1.16 | D+5.2 | 3 | 0.62× |
| FM3 Research 1998–2022 | D | 29 | +1.44 | D+6.5 | 3.3 | 0.54× |
| Cherry Communications 2012–2022 | D | 14 | +1.37 | R+0.0 | 1.6 | 0.54× |
| Public Religion Research Institute · AAPOR/Roper 2012–2020 | D | 11 | +2.01 | D+6.9 | 8.6 | 0.48× |
| McLaughlin & Associates 2000–2022 | D | 80 | +1.72 | R+5.2 | 2 | 0.44× |
How the grades work
For every historical poll we take its error on the margin and subtract the average error of the other pollsters’ polls of the same race, which removes race difficulty. A pollster’s score is the average of those differences, shrunk toward zero as if it also had ten perfectly average polls; that keeps a few lucky results from topping the table. Firms with more disclosure (transparency score) and AAPOR Transparency Initiative or Roper Center membership get a small bonus.
Grades run A+ to D. The weight column is the multiplier a pollster’s polls get in the averages (0.35× to 1.4×). “Lean” is the average signed miss when a Democrat was listed first, after at least five such polls. Data: FiveThirtyEight’s historical raw-polls file (CC BY 4.0, ABC News). Full details are in the methodology.