The biggest sample did not guarantee the right answer. The right winner did not guarantee a close margin. Open a case file to see what was measured, what happened, and what the evidence can explain.
A curated historical selection, not a definitive global ranking. Each page defines whether it concerns one poll or an aggregate, a wrong leader or a missed margin.
A huge voluntary return cannot guarantee a representative sample.
Read the evidence and analysis →The sampling and timing of a poll matter more than confidence in its winner call.
Read the evidence and analysis →Calling the winner correctly can conceal a substantial error in the victory margin.
Read the evidence and analysis →The Bradley effect is a hypothesis about mechanisms, not a complete explanation established by the miss itself.
Read the evidence and analysis →A nearly correct estimate of the winner's share can still produce a badly wrong margin.
Read the evidence and analysis →A reasonably close national popular-vote estimate can coexist with consequential errors in the states that decide the presidency.
Read the evidence and analysis →One prominent outlier can dominate the election story. Evaluate it against the result and other polls without pretending that its cause has been established.
Read the evidence and analysis →Final 1–14 days by midpoint. Latest estimate per firm/race, at least three firms, and D + R at least 90% of the recorded result.
| Contest | Category | Firms | Mean polled margin | Recorded result | Margin miss |
|---|---|---|---|---|---|
| 2018 NM-1 House Debra A. Haaland / Janice E. Arnold-Jones | House · individual districts | 3 | D +10.0 | D +22.8 | 12.84 pp |
| 2022 WA Sen Patty Murray / Tiffany Smiley | U.S. Senate | 4 | D +1.9 | D +14.5 | 12.62 pp |
| 2020 ME Sen Sara I. Gideon / Susan M. Collins | U.S. Senate | 4 | D +3.8 | R +8.6 | 12.35 pp |
| 2022 OK Sen Madison Horn / James Lankford | U.S. Senate | 3 | R +20.4 | R +32.2 | 11.77 pp |
| 2014 SD-1 House Corinna Balderramos Robinson / Kristi Noem | House · individual districts | 3 | R +21.3 | R +33.1 | 11.73 pp |
| 2014 AR Sen Mark L. Pryor / Tom Cotton | U.S. Senate | 5 | R +5.6 | R +17.1 | 11.47 pp |
| 2020 MT Pres Joe Biden / Donald Trump | President · state | 4 | R +5.0 | R +16.4 | 11.37 pp |
| 2020 KY Sen Amy McGrath / Mitch McConnell | U.S. Senate | 3 | R +8.2 | R +19.5 | 11.37 pp |
| 2022 OK Sen special Kendra Horn / Markwayne Mullin | U.S. Senate | 3 | R +15.2 | R +26.5 | 11.33 pp |
| 2000 PA Sen Ron Klink / Rick Santorum | U.S. Senate | 7 | R +18.0 | R +6.9 | 11.09 pp |
| 2020 KY Pres Joe Biden / Donald Trump | President · state | 3 | R +15.2 | R +25.9 | 10.70 pp |
| 2008 GA Sen runoff Jim Martin / Saxby Chambliss | U.S. Senate | 4 | R +4.3 | R +14.9 | 10.63 pp |
| 2016 SD Pres Hillary Rodham Clinton / Donald Trump | President · state | 3 | R +19.2 | R +29.8 | 10.56 pp |
| 2002 NM-2 House John Arthur Smith / Stevan Pearce | House · individual districts | 3 | R +2.0 | R +12.5 | 10.51 pp |
| 2012 MN-8 House Richard M. Nolan / Chip Cravaack | House · individual districts | 3 | R +1.3 | D +8.9 | 10.23 pp |
An unusual candidate pair, withdrawal, third-party contest, or sparse polling can make a ranking misleading. The declared 90% rule reduces some risks but does not prove every retained pair was the contest’s top two or that every withdrawal was identified.
Polls measure preferences; models assign win probabilities using additional assumptions. A 20% chance happening is not the same thing as a 20-point polling miss. This section evaluates measured margins, not model calibration.
Yes, as a margin miss rather than a wrong-winner upset. The 1980 presidential and 2014 Virginia Senate cases show why the size of a lead matters even when the leader ultimately wins.
