Dimanche 22 novembre 2026 Calendrier Bilan des sondages Recherche
Recherche

Comment fonctionnent nos prévisions, pays par pays

Chaque prévision du site a ses propres articles : de combien les sondages du pays se sont trompés, comment nous les corrigeons, comment les voix deviennent des sièges et ce que nous avons testé. Choisissez un pays ci-dessous ; les erreurs des sondages, de tous les pays, sont à la fin.

135 articles · environ 293k mots · Notre bilan, élection par élection

Par pays

  1. Royaume-Uni Forecasting a British general election when the polls are wrong: an overview · Measuring British polling error in England, 2015–2024 · From answers to headlines: the stage model of a published British poll 14 articles · environ 56k mots
  2. France Prévoir les élections françaises à partir de sondages imparfaits : vue d'ensemble · Mesurer l'erreur des sondages français, 1965–2024 · Comment nous lisons les sondages français : ce que nous corrigeons, ce que nous ne corrigeons pas, et pourquoi 10 articles · environ 29k mots
  3. Allemagne Forecasting the Bundestag: an overview · Measuring German polling error, 2005–2025 · How we read German polls: the 5% line and the 2023 seat law 10 articles · environ 30k mots
  4. Espagne Forecasting the Congreso from imperfect polls: an overview · Measuring Spanish polling error, 2004–2023 · Reading Spanish polls: the correction test and the CIS's raw figures 10 articles · environ 22k mots
  5. Italie Forecasting the Camera from imperfect polls: an overview · Measuring Italian polling error, 2008–2022 · Reading Italian polls: the correction and its test 11 articles · environ 19k mots
  6. Pays-Bas Forecasting the Tweede Kamer from imperfect polls: an overview · Measuring Dutch polling error, 2006–2025 · Reading Dutch polls: the correction and its test 12 articles · environ 16k mots
  7. Pologne Forecasting the Sejm from imperfect polls: an overview · Measuring Polish polling error, 2001–2023 · Reading Polish polls: why they are not corrected 12 articles · environ 16k mots
  8. Canada Prévoir la Chambre des communes à partir de sondages imparfaits : vue d'ensemble · De combien les sondages canadiens se sont trompés, 1988–2025 · Comment nous lisons les sondages canadiens : la moyenne, la correction et son test 11 articles · environ 23k mots
  9. Australie Forecasting Australia's House from imperfect polls: an overview · How wrong have Australian polls been? Primaries, the 2PP and the λ test · From primaries to seats: preference flows and the 2PP 11 articles · environ 20k mots
  10. États-Unis Forecasting the 2026 midterms: an overview · Measuring American polling error, 2000–2024 · The generic ballot: the national vote behind every race 15 articles · environ 35k mots

Quand les sondages se sont trompés

Toutes sur une page, avec l'ampleur de chaque erreur →

  1. Third time in a row: the smaller Trump miss of 2024 The final averages had Kamala Harris 0.7 points ahead and Donald Trump won the popular vote by 1.5, a 2.2-point miss that was smaller than in 2020 but the third presidential election running in which polls understated Trump. Weighting on party and recalled 2020 vote tightened the polls, but they still missed Republicans in Republican areas, the swing of Hispanic voters to Trump and new voters. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  2. Right winner, wrong size: how every poll overstated Labour's lead in 2024 All 19 final polls listed by the British Polling Council overstated Labour's lead, averaging 17.2 points against an actual 10.3, the largest overestimate of Labour since 1979. Pollsters' reviews pointed to too few older and loyal Conservatives in their samples, high refusal rates, over-engaged online panels and some late switching away from Labour. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  3. Justes sur les voix, faux sur les sièges : le front républicain et l'erreur des projections de 2024 Les sondages de premier tour étaient proches du résultat (bloc RN 36,1 contre 33,2, NFP 28,4 contre 28,0), mais toutes les dernières projections en sièges faisaient du bloc RN le premier groupe avec 170 à 250 sièges, et il a terminé troisième avec 142. Plus de 200 désistements entre les deux tours et des électeurs inhabituellement disciplinés ont rompu les reports sur lesquels reposaient les projections. 9 min de lecture · mis à jour le 2 novembre 2026
  4. Wilders' last-minute landslide: how the 2023 Dutch polls missed the PVV by nine seats The final polls showed a three-way tie, with the PVV on 27 to 29 seats; it won 37 seats and 23.5% as undecided and tactical voters on the right broke for Wilders in the last two days. All three pollsters missed in the same direction, the mark of a late swing rather than one firm's method. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  5. The majority that wasn't: Spain's 2023 polling miss and the CIS outlier The final polls had the PP 5.7 points ahead of the PSOE and the PP and Vox close to a majority; the PP won by 1.4 points and the right finished six seats short. Every private pollster understated the PSOE, and only the CIS, which named the wrong winner, did not. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  6. The biggest miss in 40 years: Biden's shrinking lead in 2020 The final averages had Joe Biden 7.9 points ahead and he won the popular vote by 4.5, the largest national polling error in 40 years by AAPOR's count, with state polls missing by even more. AAPOR traced at least part of it to nonresponse: the people who answered polls were more Democratic than the people who voted, in ways demographic weighting could not fix. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  7. The miracle that was a sampling problem: Australia's 2019 polling failure All five final polls had Labor winning the two-party-preferred vote 51.4 to 48.6; the Coalition won 51.5 to 48.5. The industry inquiry blamed samples that held too many engaged, highly educated voters and were never weighted to correct it. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  8. Fixing 2015 too hard: turnout models and the majority that vanished in 2017 The final polls had the Conservatives 8 points ahead and they won by 2.5, losing their majority; the Conservative share was right and Labour's 5 points too low. Raw poll data were close to the result, and the turnout models pollsters had added after 2015 pushed the headlines towards the Conservatives, while Survation and YouGov's MRP model came close. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  9. Rutte pulls away, Wilders fades: the left-leaning panels behind the 2017 Dutch polling miss The final polls had the VVD only about five seats ahead of the PVV; Rutte won by 13 seats and 8.2 points. Every pollster understated the VVD, and the pollsters' own re-interviews found panels that leaned left plus a late move to Rutte during the row with Turkey. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  10. The Blue Wall that wasn't: state polls and the 2016 miss National polls had Hillary Clinton about 4 points ahead and she won the popular vote by 2.1, a normal-sized miss, but state polls overstated her by 5 to 7 points in Wisconsin, Michigan and Pennsylvania, all of which Donald Trump won. AAPOR blamed late deciders breaking for Trump and state polls that did not weight by education. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  11. The sorpasso that never came: Spain's June 2016 polling miss All 22 final polls had Unidos Podemos second, ahead of the PSOE, and the PP on under 30%; the coalition finished third with 21.2% and the PP won 33.0%. The polls counted the December voters of Podemos and Izquierda Unida as if they would all back the joint list, and more than a million did not. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  12. Ten points clear, then second: how the polls missed Duda's first-round win in 2015 Every final poll had Bronisław Komorowski about ten points ahead of Andrzej Duda; Duda came first by one point and Paweł Kukiz beat his polls by five. Turnout of 49% and a late break away from the incumbent undid polls that assumed a much larger, steadier electorate. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  13. The hung parliament that never came: unrepresentative samples and the 2015 polling miss Every BPC member's final poll had the Conservatives and Labour within a point or two of each other, and the Conservatives won Great Britain by 6.6 points and a majority. The Sturgis inquiry found the samples held too many Labour supporters and too many politically engaged people, and ruled out shy Tories and late swing as major causes. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  14. Les jeunes électeurs restés chez eux : le raté des sondages en Colombie-Britannique en 2013 Tous les sondeurs donnaient le NPD en tête, de 7 points en moyenne ; les libéraux de Christy Clark l'ont emporté par 44,1 % contre 39,7 %. Les bilans des firmes ont montré qu'elles avaient pondéré les jeunes, favorables au NPD, selon leur poids dans la population et non parmi les votants, et qu'elles avaient manqué l'avantage de participation des libéraux. 8 min de lecture · mis à jour le 2 novembre 2026
  15. Grillo's tsunami: how Italy's 2013 polls missed the Five Star surge by ten points The last polls before the blackout had Bersani's centre-left about 5 points clear and the M5S on 15%; the centre-left won the Chamber by 0.4 points and the M5S took 25.6%, the largest single list. Every pollster understated the M5S and overstated the PD, through a new party with no past vote to weight on, reluctant centre-left switchers and a late surge inside the 16-day blackout. 6 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  16. L'avance du Wildrose disparue en une nuit : le raté des sondages albertains de 2012 Les derniers sondages donnaient en moyenne 41 % au Wildrose et 34 % aux progressistes-conservateurs ; les PC l'ont emporté par 44,0 % contre 34,3 % et ont conservé une majorité de 61 sièges. Les sondeurs ont invoqué un virage de dernière minute des indécis et d'électeurs de centre gauche vers les PC, après la fin de la plupart des sondages, aggravé par une participation en hausse de 14 points. 7 min de lecture · mis à jour le 2 novembre 2026
  17. Six points lost in the last week: Merkel's vanishing lead and Germany's 2005 polling miss The final polls had the CDU/CSU on 41.5% and eight points clear of the SPD; it won 35.2% and finished less than a point ahead. Every institute overstated the Union by 5.8 to 6.8 points, and their own analyses found a late swing of Union-leaning voters to the FDP that the broadcasters' polls were not allowed to publish. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais
  18. Treize pour cent dans les sondages, deuxième le soir du vote : Le Pen et l'erreur des sondages du 21 avril 2002 Les derniers sondages donnaient Jospin à 17,8 % et Le Pen à 13,0 % ; Le Pen a obtenu 16,9 % et éliminé Jospin (16,2 %) du second tour. Tous les instituts ont sous-estimé l'extrême droite et surestimé les petits candidats de la gauche sortante, et la Commission des sondages a mis en cause les échantillons téléphoniques, des redressements politiques trop prudents et le lissage des chiffres publiés. 8 min de lecture · mis à jour le 2 novembre 2026
  19. Shy Tories and a late swing: how the 1992 polls handed Kinnock a lead he never had The final polls put Labour about a point ahead and the Conservatives won Great Britain by 7.6 points, the worst miss on the lead since polling began. The Market Research Society inquiry blamed late swing, quota samples that did not match the electorate and Conservatives who would not say how they would vote. 7 min de lecture · mis à jour le 2 novembre 2026 · en anglais

À voir aussi