How close the UK's final polls have come to the result at past elections, party by party and pollster by pollster, and what that means for today's average (Labour 28.4, Reform UK 21.9, Conservative 21.7, Lib Dem 9.9).
Every election, all ten countries: polls on the eve vs the result · which countries poll best · pollster leaderboard
Each card is one party: every dot is one election's final England polls against the result. Red means the polls overstated the party, blue that they understated it. The shaded band is the 80% range we allow for today.
Why we measure the polls in England
Every poll you see in the news is a Great Britain figure. We work with England's figure from the same poll, taken from its published tables. That is what lets us predict the next miss.
- One contest, not three. England, Scotland and Wales have different parties. The SNP and Plaid Cymru only stand in Scotland and Wales, so a GB share mixes three separate races. In 2015 Labour's vote fell from 42% to 24% in Scotland but rose in England. The GB figure hid both moves. In England every voter chooses between the same parties, so a poll's error there is a like-for-like miss.
- It is where the seats are. 543 of the 650 seats are in England. We forecast them from England's vote shares, by swing from the 2024 England result. So the error that moves our seat forecast is the England error, and that is the one we measure and correct.
- It is how we correct each poll. For each pollster we take the England figure from its raw answers, then take off part of the average amount raw England figures have missed by before. The past miss and the poll it corrects are measured the same way, so the correction applies like for like. A GB miss would carry Scottish swings (the 2015 collapse, for example) into English seats.
- It costs little precision. About six in seven people in a GB poll live in England, so the England figure rests on nearly the whole sample. Scotland and Wales get their own polls and their own record of polling error (Scotland & Wales).
The catch: England figures and raw tables are only published consistently from 2015, so this record covers four elections. For the longer history we use Britain-wide final polls back to 1945 (Long view). They show which way the polls tend to miss, but they can't feed the correction.
Pollsters in today's average, and their record
Lead error is how far a pollster's final poll got Labour's lead over the Conservatives wrong: +3 means it was 3 points too good for Labour, −3 too good for the Conservatives. Today's difference is the pollster's latest England figure minus the average: it shows which way each one leans now.
| Pollster | Today (England) | vs the average | Lab–Con lead error at the final poll | Avg miss | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Lab | Con | Ref | Lab | Con | Ref | 2015 | 2017 | 2019 | 2024 | ||
| Opinium | 30 | 22 | 23 | +0.8 | -1.3 | +0.7 | +7.2* | -3.9 | +2.0 | +9.8 | 1.6 |
| Survation | 31 | 22 | 25 | +1.4 | -0.6 | +2.4 | +8.0 | +1.2 | +1.1 | – | 1.5 |
| YouGov | 27 | 24 | 22 | -2.7 | +0.9 | -0.2 | +6.2 | – | – | +6.9 | 2.0 |
| Find Out Now | 30 | 21 | 22 | -0.0 | -2.2 | -0.4 | – | – | – | – | – |
| More in Common | 30 | 26 | 20 | +0.5 | +3.2 | -2.6 | – | – | – | – | – |
|
Ipsos
not in current average
|
– | -8.3 | +1.8 | +9.3 | 1.7 | ||||||
|
Savanta
not in current average
|
+7.2* | -6.8 | +8.4 | +11.0 | 2.8 | ||||||
|
JL Partners
not in current average
|
– | – | – | +5.9 | 2.1 | ||||||
|
Norstat
not in current average
|
– | – | – | +4.0 | 1.7 | ||||||
|
Qriously
not in current average
|
– | – | -0.7 | – | 1.5 | ||||||
|
BMG Research
not in current average
|
+6.2* | – | – | – | – | ||||||
|
Deltapoll
not in current average
|
– | – | +1.9* | +9.1* | – | ||||||
|
ICM Unlimited
not in current average
|
+9.2* | – | – | – | – | ||||||
|
Lord Ashcroft Polls
not in current average
|
+7.6* | – | – | – | – | ||||||
|
Number Cruncher Politics
not in current average
|
– | – | +3.3* | +6.8* | – | ||||||
* Sensitivity-only poll (not on the pollster's headline basis, rebuilt with estimated bases, or taken from a secondary source), so it is left out of the averages. 5 pollsters are in today's England average; 3 of them have a checked final-poll record. Each pollster's full record and our rating are on the pollster ratings page. Want to see what happens if you only trust some of them? Try the pollster rows on the forecast or the what-if tool.
Positive = the polls overstated the party; negative = understated.
2015 general election
| Party | Polls said | Result | Error | Direction | Polls |
|---|---|---|---|---|---|
| Conservative | 35.3 | 40.9 | -5.7 | Underestimated | 2 |
| Labour | 33.2 | 31.7 | +1.5 | Overestimated | 2 |
| Lib Dem | 10.0 | 8.2 | +1.8 | Overestimated | 2 |
| Green | 5.1 | 4.2 | +0.9 | Overestimated | 2 |
| UKIP | 15.0 | 14.1 | +0.8 | Overestimated | 2 |
| Others | 1.5 | 0.8 | +0.7 | Overestimated | 2 |
2017 general election
| Party | Polls said | Result | Error | Direction | Polls |
|---|---|---|---|---|---|
| Conservative | 46.1 | 45.5 | +0.6 | Overestimated | 4 |
| Labour | 38.1 | 42.0 | -3.9 | Underestimated | 4 |
| Lib Dem | 8.6 | 7.8 | +0.7 | Overestimated | 4 |
| Green | 2.5 | 1.9 | +0.6 | Overestimated | 4 |
| UKIP | 4.2 | 2.1 | +2.1 | Overestimated | 4 |
| Others | 0.6 | 0.8 | -0.2 | Underestimated | 4 |
2019 general election
| Party | Polls said | Result | Error | Direction | Polls |
|---|---|---|---|---|---|
| Conservative | 45.6 | 47.3 | -1.6 | Underestimated | 5 |
| Labour | 34.8 | 33.9 | +0.9 | Overestimated | 5 |
| Lib Dem | 11.9 | 12.4 | -0.6 | Underestimated | 5 |
| Green | 3.1 | 3.0 | +0.1 | Overestimated | 5 |
| Brexit Party | 2.8 | 2.0 | +0.8 | Overestimated | 5 |
| Others | 1.8 | 1.3 | +0.5 | Overestimated | 5 |
2024 general election
| Party | Polls said | Result | Error | Direction | Polls |
|---|---|---|---|---|---|
| Conservative | 22.3 | 25.9 | -3.5 | Underestimated | 6 |
| Labour | 38.8 | 34.5 | +4.3 | Overestimated | 6 |
| Lib Dem | 12.3 | 13.2 | -0.9 | Underestimated | 6 |
| Reform UK | 17.3 | 15.4 | +2.0 | Overestimated | 6 |
| Green | 6.8 | 7.3 | -0.6 | Underestimated | 6 |
| Others | 2.4 | 3.8 | -1.3 | Underestimated | 6 |
Labour–Conservative lead
| Election | Pair | Polls | Result | Error |
|---|---|---|---|---|
| 2015 | LAB-CON | -2.1 | -9.2 | +7.1 |
| 2017 | LAB-CON | -8.0 | -3.5 | -4.5 |
| 2019 | LAB-CON | -10.8 | -13.3 | +2.5 |
| 2024 | LAB-CON | 16.4 | 8.6 | +7.8 |
Does correcting the polls help?
We tested several ways of adjusting the polls for past error, each time using only earlier elections to predict the next one. A correction is only worth using if it beats the raw polls in every election tested.
| Model | Variant | Setting | Avg. error | vs raw | Elections beaten | Consistent? |
|---|---|---|---|---|---|---|
| M3_shrunk | PartyFamilyHistory | 2.000 | 1.49 | -0.05 | 2 / 3 | No |
| M3_shrunk | PartyFamilyHistory | 4.000 | 1.49 | -0.04 | 2 / 3 | No |
| M0_raw | PartyFamilyHistory | 0.000 | 1.54 | 0.00 | 0 / 3 | No |
| M3_shrunk | PartyFamilyHistory | 1.000 | 1.55 | +0.02 | 2 / 3 | No |
| M2_decay | PartyFamilyHistory | 0.250 | 1.98 | +0.44 | 1 / 3 | No |
| M2_decay | PartyFamilyHistory | 0.100 | 1.99 | +0.46 | 1 / 3 | No |
| M1_mean | PartyFamilyHistory | 0.000 | 1.99 | +0.46 | 1 / 3 | No |
| M2_decay | PartyFamilyHistory | 0.500 | 2.02 | +0.48 | 1 / 3 | No |
| M4_last | PartyFamilyHistory | 0.000 | 2.12 | +0.59 | 1 / 3 | No |
| M0_raw | PartyFamilyHistory_no2015 | 0.000 | 1.51 | 0.00 | 0 / 2 | No |
| M3_shrunk | PartyFamilyHistory_no2015 | 4.000 | 1.57 | +0.05 | 1 / 2 | No |
| M3_shrunk | PartyFamilyHistory_no2015 | 2.000 | 1.61 | +0.10 | 1 / 2 | No |
| M3_shrunk | PartyFamilyHistory_no2015 | 1.000 | 1.73 | +0.21 | 1 / 2 | No |
| M4_last | PartyFamilyHistory_no2015 | 0.000 | 1.76 | +0.25 | 1 / 2 | No |
| M2_decay | PartyFamilyHistory_no2015 | 0.500 | 1.93 | +0.41 | 1 / 2 | No |
| M2_decay | PartyFamilyHistory_no2015 | 0.250 | 2.02 | +0.51 | 1 / 2 | No |
| M2_decay | PartyFamilyHistory_no2015 | 0.100 | 2.09 | +0.58 | 1 / 2 | No |
| M1_mean | PartyFamilyHistory_no2015 | 0.000 | 2.14 | +0.62 | 1 / 2 | No |
| M0_raw | StrictPartyHistory | 0.000 | 1.54 | 0.00 | 0 / 3 | No |
| M3_shrunk | StrictPartyHistory | 4.000 | 1.56 | +0.03 | 1 / 3 | No |
| M3_shrunk | StrictPartyHistory | 2.000 | 1.59 | +0.05 | 1 / 3 | No |
| M3_shrunk | StrictPartyHistory | 1.000 | 1.65 | +0.11 | 1 / 3 | No |
| M1_mean | StrictPartyHistory | 0.000 | 1.99 | +0.46 | 1 / 3 | No |
| M2_decay | StrictPartyHistory | 0.100 | 2.00 | +0.47 | 1 / 3 | No |
| M2_decay | StrictPartyHistory | 0.250 | 2.04 | +0.51 | 1 / 3 | No |
| M2_decay | StrictPartyHistory | 0.500 | 2.08 | +0.55 | 1 / 3 | No |
| M4_last | StrictPartyHistory | 0.000 | 2.15 | +0.62 | 1 / 3 | No |
Correcting the published figures by their past error did not help reliably. Correcting the pollsters' raw answers did, and that is our headline forecast (our projection); ranges come from the size of past errors. See Measuring polling error.
Range used for each party
| Party | 80% range (±) | How it was set |
|---|---|---|
| Conservative | 4.4 | No correction. Band = 1.2816 x RMS of 4 historical England industry errors (2015: -5.67, 2017: +0.59, 2019: -1.63, 2024: -3.54). |
| Green | 0.8 | No correction. Band = 1.2816 x RMS of 4 historical England industry errors (2015: +0.86, 2017: +0.59, 2019: +0.06, 2024: -0.59). |
| Labour | 3.9 | No correction. Band = 1.2816 x RMS of 4 historical England industry errors (2015: +1.48, 2017: -3.87, 2019: +0.89, 2024: +4.28). |
| Lib Dem | 1.4 | No correction. Band = 1.2816 x RMS of 4 historical England industry errors (2015: +1.82, 2017: +0.74, 2019: -0.57, 2024: -0.87). |
| Others | 1.0 | No correction. Band = 1.2816 x RMS of 4 historical England industry errors (2015: +0.67, 2017: -0.16, 2019: +0.49, 2024: -1.32). |
| Reform UK | 2.0 | No correction. Band = 1.2816 x RMS of 4 historical England industry errors (2015: +0.83, 2017: +2.11, 2019: +0.75, 2024: +1.96). |
Scotland and Wales
Poll minus result, in points · each nation's own pollsEngland's record can't simply be copied to Scotland and Wales: the SNP and Plaid Cymru stand there, and the Scottish and Welsh polls are separate surveys. So each nation has its own record, built from Scotland-only and Wales-only polls: each pollster's final one before polling day, checked against its own tables. The Scotland and Wales columns of the GB polls are shown alongside for comparison. They are far smaller samples, and miss by roughly twice as much.
Scotland
| Election | Polls | SNP | LAB | CON | LD | GRN | REF | Average miss | GB polls' Scotland column |
|---|---|---|---|---|---|---|---|---|---|
| 2015 | 3 | +0.3 | 0.0 | +0.6 | −1.6 | +0.4 | −0.3 | 2.0 | 2.8 (3) |
| 2017 | 3 | +4.1 | −0.9 | −2.9 | −1.2 | +0.8 | −0.2 | 2.5 | 4.7 (5) |
| 2019 | 3 | −1.3 | −1.6 | +2.2 | +0.6 | +0.2 | −0.4 | 2.0 | 3.5 (6) |
| 2024 | 6 | +2.4 | −0.5 | +1.0 | −1.6 | −0.6 | −0.7 | 1.9 | 4.5 (6) |
Does correcting Scotland's polls for their average past miss help? Tested election by election using only the other elections: average miss 2.25 corrected against 2.10 as published (2015: 2.1 vs 2.0, 2017: 2.4 vs 2.5, 2019: 2.8 vs 2.0, 2024: 1.7 vs 1.9). It didn't help: the misses don't repeat, so our forecast leaves Scotland's polls as they are and uses the past misses only for the ranges and the "polls as wrong as in…" options.
The Scotland polls behind this
| Election | Pollster | Fieldwork end | Sample | SNP | LAB | CON | LD | GRN | REF | Kind |
|---|---|---|---|---|---|---|---|---|---|---|
| 2015 | Ipsos MORI | 27 Apr | 1071 | +4.0 | −4.3 | +2.1 | −2.6 | +0.6 | −0.6 | Scotland poll |
| 2015 | Survation | 6 May | 1660 | −1.1 | +0.5 | +0.6 | −1.6 | +1.0 | +0.4 | Scotland poll |
| 2015 | YouGov | 6 May | 1351 | −2.0 | +3.7 | −0.9 | −0.6 | −0.4 | −0.6 | Scotland poll |
| 2015 | Lord Ashcroft Polls | 6 May | 167 | +4.0 | −0.3 | −3.9 | −1.6 | +0.6 | −0.6 | GB poll's column |
| 2015 | Populus | 7 May | 224 | −8.0 | +5.7 | −1.9 | −0.6 | +2.6 | +1.4 | GB poll's column |
| 2015 | Survation | 6 May | 342 | +3.5 | −1.5 | −0.7 | −2.2 | −0.6 | +0.9 | GB poll's column |
| 2017 | Ipsos MORI | 27 May | 1016 | +6.1 | −2.1 | −3.6 | −1.8 | – | – | Scotland poll |
| 2017 | Survation | 7 Jun | 1001 | +2.1 | +1.5 | −2.4 | −1.0 | – | – | Scotland poll |
| 2017 | YouGov | 5 Jun | 1093 | +4.1 | −2.1 | −2.6 | −0.8 | +0.8 | −0.2 | Scotland poll |
| 2017 | ComRes | 7 Jun | 108 | +8.1 | −11.1 | −1.6 | +2.2 | −0.2 | +0.8 | GB poll's column |
| 2017 | Ipsos MORI | 7 Jun | 71 | +10.1 | −3.1 | −6.6 | −2.8 | +1.8 | +0.8 | GB poll's column |
| 2017 | Opinium | 6 Jun | 189 | +12.1 | −10.1 | −1.6 | −1.8 | +0.8 | −0.2 | GB poll's column |
| 2017 | Survation | 7 Jun | 205 | +4.0 | +2.2 | −4.0 | −2.1 | −0.2 | −0.2 | GB poll's column |
| 2017 | YouGov | 7 Jun | 196 | +4.1 | −5.1 | −0.6 | +0.2 | −0.2 | −0.2 | GB poll's column |
| 2019 | Ipsos MORI | 25 Nov | 1046 | −1.0 | −2.6 | +0.9 | +1.4 | +1.0 | −0.6 | Scotland poll |
| 2019 | Survation | 11 Dec | 1004 | −1.8 | +1.2 | +2.8 | −2.2 | −0.2 | −0.1 | Scotland poll |
| 2019 | YouGov | 3 Dec | 1002 | −1.0 | −3.6 | +2.9 | +2.4 | 0.0 | −0.6 | Scotland poll |
| 2019 | Ipsos MORI | 11 Dec | 147 | +3.0 | −1.6 | −2.1 | −1.6 | +1.0 | +0.4 | GB poll's column |
| 2019 | Number Cruncher Politics | 10 Dec | 89 | +2.0 | −5.6 | +0.9 | −0.6 | +1.0 | +1.4 | GB poll's column |
| 2019 | Opinium | 11 Dec | 161 | −1.8 | −1.3 | +7.8 | −5.2 | +0.8 | −0.1 | GB poll's column |
| 2019 | Qriously | 8 Dec | 88 | −14.1 | +8.7 | −0.8 | +2.4 | −1.0 | −0.6 | GB poll's column |
| 2019 | Savanta ComRes | 10 Dec | 159 | −5.0 | −4.6 | +5.9 | +2.4 | −1.0 | +2.4 | GB poll's column |
| 2019 | Survation | 11 Dec | 157 | +0.6 | −3.8 | +2.3 | 0.0 | +0.1 | +0.8 | GB poll's column |
| 2024 | Ipsos | 9 Jun | 1150 | +6.0 | +0.7 | +0.3 | −4.7 | −0.8 | −3.0 | Scotland poll |
| 2024 | More in Common | 28 Jun | 1008 | −0.1 | −0.5 | +2.8 | −0.8 | −1.4 | +0.3 | Scotland poll |
| 2024 | Opinium | 10 Jun | 1017 | +4.2 | −0.7 | +1.3 | −1.6 | −0.3 | −2.3 | Scotland poll |
| 2024 | Savanta | 2 Jul | 1083 | +4.0 | −4.3 | +2.3 | −0.7 | −0.8 | −1.0 | Scotland poll |
| 2024 | Survation | 25 Jun | 1022 | +1.1 | +1.7 | +1.0 | −2.9 | −1.0 | +0.6 | Scotland poll |
| 2024 | YouGov | 25 Jun | 1059 | −1.0 | −0.3 | −1.7 | +1.3 | +1.2 | +1.0 | Scotland poll |
| 2024 | Ipsos | 3 Jul | 142 | +22.0 | −14.3 | −3.7 | −5.7 | +3.2 | −3.0 | GB poll's column |
| 2024 | J.L. Partners | 3 Jul | 124 | −3.0 | +1.7 | +3.3 | +0.3 | −2.8 | 0.0 | GB poll's column |
| 2024 | Norstat | 3 Jul | 211 | +3.0 | −3.3 | +2.3 | −2.7 | −1.8 | +3.0 | GB poll's column |
| 2024 | Number Cruncher Politics | 3 Jul | 215 | −7.0 | +10.7 | −6.7 | +0.3 | +5.2 | 0.0 | GB poll's column |
| 2024 | Opinium | 3 Jul | 155 | +0.6 | +5.1 | −0.5 | −5.0 | −2.0 | +3.3 | GB poll's column |
| 2024 | Savanta | 3 Jul | 140 | −4.0 | +0.7 | +2.3 | +0.3 | −3.8 | −1.0 | GB poll's column |
Sample: people asked for a Scotland poll; weighted Scotland base for a GB poll's column. A dash means the poll didn't print that party (it sits in Other). Not included because the tables couldn't be reached: Panelbase (2015, 2017, 2019), TNS (2015), BMG (2017), Norstat and Redfield & Wilton (2024).
Wales
| Election | Polls | LAB | CON | PC | LD | REF | GRN | Average miss | GB polls' Wales column |
|---|---|---|---|---|---|---|---|---|---|
| 2015 | 1 | +2.1 | −2.2 | +0.9 | +1.5 | −1.6 | −0.6 | 1.7 | 3.1 (2) |
| 2017 | 1 | −3.0 | +0.4 | −1.4 | +0.5 | +3.0 | −0.3 | 1.7 | 6.2 (4) |
| 2019 | 1 | −0.9 | +0.9 | +0.1 | 0.0 | −0.4 | 0.0 | 0.5 | 4.6 (6) |
| 2024 | 1 | +8.0 | −0.2 | −2.8 | −1.5 | −4.0 | −0.7 | 3.3 | 4.5 (6) |
Does correcting Wales's polls for their average past miss help? Tested election by election using only the other elections: average miss 2.31 corrected against 1.77 as published (2015: 1.7 vs 1.7, 2017: 2.7 vs 1.7, 2019: 1.3 vs 0.5, 2024: 3.6 vs 3.3). It didn't help: the misses don't repeat, so our forecast leaves Wales's polls as they are and uses the past misses only for the ranges and the "polls as wrong as in…" options.
The Wales polls behind this
| Election | Pollster | Fieldwork end | Sample | LAB | CON | PC | LD | REF | GRN | Kind |
|---|---|---|---|---|---|---|---|---|---|---|
| 2015 | YouGov | 6 May | 1202 | +2.1 | −2.2 | +0.9 | +1.5 | −1.6 | −0.6 | Wales poll |
| 2015 | Lord Ashcroft Polls | 6 May | 85 | −0.7 | +0.5 | +4.7 | −2.9 | −4.0 | +3.5 | GB poll's column |
| 2015 | Survation | 6 May | 192 | −5.9 | −0.6 | −3.8 | +3.3 | +4.3 | +2.4 | GB poll's column |
| 2017 | YouGov | 7 Jun | 1074 | −3.0 | +0.4 | −1.4 | +0.5 | +3.0 | −0.3 | Wales poll |
| 2017 | ComRes | 7 Jun | 62 | −11.0 | +6.4 | +3.6 | +0.5 | +1.0 | −0.3 | GB poll's column |
| 2017 | Ipsos MORI | 7 Jun | 45 | +10.0 | −7.9 | −10.4 | −2.0 | −2.0 | −0.3 | GB poll's column |
| 2017 | Opinium | 6 Jun | 104 | −10.0 | +2.4 | +2.6 | +0.5 | +6.0 | −0.3 | GB poll's column |
| 2017 | Survation | 7 Jun | 119 | −12.2 | −10.7 | +23.1 | −1.4 | +0.7 | −0.3 | GB poll's column |
| 2019 | YouGov | 9 Dec | 1020 | −0.9 | +0.9 | +0.1 | 0.0 | −0.4 | 0.0 | Wales poll |
| 2019 | Ipsos MORI | 11 Dec | 96 | −4.7 | +4.6 | +2.2 | +1.7 | −2.1 | −1.0 | GB poll's column |
| 2019 | Number Cruncher Politics | 10 Dec | 52 | −15.9 | +5.9 | +0.1 | +8.0 | −1.4 | +3.0 | GB poll's column |
| 2019 | Opinium | 11 Dec | 106 | −6.3 | +3.7 | −4.6 | +5.8 | +1.3 | −1.0 | GB poll's column |
| 2019 | Qriously | 8 Dec | 53 | −10.4 | +14.8 | −6.0 | −1.4 | +1.5 | +0.6 | GB poll's column |
| 2019 | Savanta ComRes | 10 Dec | 87 | −1.9 | −4.1 | +0.1 | +2.0 | +3.6 | 0.0 | GB poll's column |
| 2019 | Survation | 11 Dec | 145 | −4.2 | −4.0 | +7.7 | −4.1 | +4.5 | 0.0 | GB poll's column |
| 2024 | YouGov | 3 Jun | 1066 | +8.0 | −0.2 | −2.8 | −1.5 | −4.0 | −0.7 | Wales poll |
| 2024 | Ipsos | 3 Jul | 82 | +4.5 | −5.3 | +2.4 | −2.2 | +4.5 | −3.2 | GB poll's column |
| 2024 | J.L. Partners | 3 Jul | 59 | −5.0 | −2.2 | −4.8 | +5.5 | +13.0 | −3.7 | GB poll's column |
| 2024 | Norstat | 3 Jul | 109 | 0.0 | +8.8 | −6.8 | +1.5 | −1.0 | −0.7 | GB poll's column |
| 2024 | Number Cruncher Politics | 3 Jul | 120 | +8.0 | −0.2 | −3.8 | −1.5 | −1.0 | −1.7 | GB poll's column |
| 2024 | Opinium | 3 Jul | 107 | +11.2 | −8.6 | −9.0 | +3.4 | +4.1 | +1.0 | GB poll's column |
| 2024 | Savanta | 3 Jul | 79 | +6.0 | −1.2 | +5.2 | −4.5 | −1.0 | −3.7 | GB poll's column |
Sample: people asked for a Wales poll; weighted Wales base for a GB poll's column. A dash means the poll didn't print that party (it sits in Other). Not included because the tables couldn't be reached: ICM (2015), and More in Common, Redfield & Wilton and Savanta (2024). Wales has one Wales-only poll per election (YouGov); its 2024 poll was taken a month before polling day, so part of its Labour miss is late movement.
Our England record starts in 2015, when pollsters began publishing England figures and raw tables. Before that, all we have is each company's final Britain-wide poll. Here is every one of them since 1945, against the GB result: 151 final polls from 44 companies across 22 general elections. It can't feed our correction, which needs England figures and raw answers, but it does show which way the polls have tended to miss.
Error on Labour's lead at every general election since 1945
Each dot is one company's final poll: its Labour lead over the Conservatives minus the real lead in Britain. Above the line, too good for Labour; below, too good for the Conservatives. The bar is the average.
By election
| Election | Final polls | Con | Lab | Lib/LD | Lab lead | Avg miss | Closest | Furthest |
|---|---|---|---|---|---|---|---|---|
| 1945 | 1 | +1.7 | −1.8 | +1.3 | −3.5 | 1.6 | Gallup | |
| 1950 | 3 | +1.5 | −3.0 | +1.9 | −4.5 | 2.1 | Gallup | Daily Mail (own poll) |
| 1951 | 3 | +2.0 | −4.1 | +1.8 | −6.1 | 2.6 | Gallup | Research Services |
| 1955 | 1 | +1.7 | +0.1 | −1.3 | −1.6 | 1.0 | Gallup | |
| 1959 | 4 | −0.2 | +0.9 | −0.4 | +1.0 | 1.0 | Daily Express (own poll) | Gallup |
| 1964 | 4 | +1.9 | +1.2 | −2.7 | −0.7 | 2.1 | Daily Express (own poll) | Research Services |
| 1966 | 4 | −1.2 | +2.6 | −0.8 | +3.8 | 1.6 | Research Services | Daily Express (own poll) |
| 1970 | 5 | −2.2 | +4.3 | −1.1 | +6.5 | 2.6 | ORC | Marplan |
| 1974 Feb | 6 | −0.6 | −2.2 | +2.9 | −1.6 | 2.2 | Gallup | Marplan |
| 1974 Oct | 6 | −2.6 | +2.4 | +0.6 | +5.0 | 1.9 | Gallup | NOP |
| 1979 | 6 | +0.9 | +0.3 | −1.0 | −0.7 | 1.0 | ORC | Research Services |
| 1983 | 6 | +2.4 | −2.7 | +0.5 | −5.1 | 1.9 | Ipsos | ASL |
| 1987 | 6 | −1.0 | +2.7 | −1.5 | +3.6 | 1.8 | Ipsos | Harris |
| 1992 | 8 | −4.8 | +4.0 | +1.0 | +8.8 | 3.5 | Nielsen | ASL |
| 1997 | 5 | −0.7 | +2.3 | −0.8 | +3.0 | 2.2 | ICM | NOP |
| 2001 | 6 | −1.1 | +3.2 | −1.5 | +4.3 | 2.0 | ICM | NOP |
| 2005 | 8 | −1.1 | +1.8 | −0.1 | +2.8 | 1.2 | NOP | ComRes / Savanta |
| 2010 | 12 | −1.8 | −2.7 | +4.3 | −0.9 | 3.0 | RNB | OnePoll |
| 2015 | 12 | −3.6 | +1.8 | +0.8 | +5.4 | 2.4 | Survation | Panelbase |
| 2017 | 13 | −0.5 | −4.7 | +0.2 | −4.2 | 2.4 | Survation | BMG |
| 2019 | 12 | −1.5 | +0.5 | +0.1 | +2.0 | 1.3 | Opinium | ComRes / Savanta |
| 2024 | 20 | −3.3 | +4.2 | −1.5 | +7.5 | 3.0 | Norstat | Focaldata |
Mean error of the final polls against the GB result, in points: + means the polls overstated the party. Greens and UKIP / Brexit Party / Reform are in the list below from 2015, the first election PollBase tracks them in every poll.
By polling company
| Company | Elections | Span | Avg lead error | Too good for Labour | Avg miss |
|---|---|---|---|---|---|
| Gallup | 16 | 1945–2001 | +1.1 | 8 of 16 | 1.5 |
| NOP | 13 | 1959–2005 | +3.3 | 11 of 13 | 2.3 |
| Ipsos | 12 | 1979–2024 | +2.3 | 8 of 12 | 1.6 |
| Harris | 9 | 1970–2010 | +2.3 | 7 of 9 | 2.2 |
| ICM | 8 | 1992–2019 | +1.5 | 5 of 8 | 2.0 |
| YouGov | 7 | 2001–2024 | +1.9 | 6 of 7 | 2.0 |
| Marplan | 6 | 1970–1987 | +2.9 | 4 of 6 | 2.4 |
| ComRes / Savanta | 6 | 2005–2024 | +2.8 | 4 of 6 | 2.5 |
| Daily Express (own poll) | 5 | 1950–1966 | −0.6 | 2 of 5 | 1.9 |
| Kantar / Verian | 5 | 2010–2024 | +1.7 | 3 of 5 | 2.0 |
| Opinium | 5 | 2010–2024 | +1.9 | 2 of 5 | 2.0 |
| Research Services | 4 | 1951–1979 | −3.3 | 1 of 4 | 2.7 |
| ORC | 4 | 1970–1979 | +0.8 | 3 of 4 | 1.3 |
| BMG | 4 | 2015–2024 | +1.4 | 3 of 4 | 2.9 |
| Survation | 4 | 2015–2024 | +2.6 | 4 of 4 | 1.5 |
| ASL | 3 | 1983–1992 | +1.9 | 2 of 3 | 3.4 |
| Populus | 3 | 2005–2015 | +2.6 | 2 of 3 | 2.0 |
| Panelbase | 3 | 2015–2019 | +1.9 | 2 of 3 | 2.0 |
| Business Decisions | 2 | 1974 Feb–1974 Oct | +2.2 | 2 of 2 | 1.8 |
| BPIX | 2 | 2005–2010 | +2.2 | 2 of 2 | 2.8 |
| Lord Ashcroft | 2 | 2015–2024 | +7.6 | 2 of 2 | 3.1 |
| SurveyMonkey | 2 | 2015–2017 | −0.6 | 1 of 2 | 2.3 |
| Norstat | 2 | 2017–2024 | +0.6 | 1 of 2 | 2.5 |
| Qriously | 2 | 2017–2019 | +1.6 | 1 of 2 | 1.6 |
| Deltapoll | 2 | 2019–2024 | +4.2 | 2 of 2 | 2.3 |
| Number Cruncher | 2 | 2019–2024 | +4.8 | 2 of 2 | 1.9 |
| Daily Mail (own poll) | 1 | 1950 | −6.8 | 0 of 1 | 3.2 |
| Forecasting Statistics | 1 | 1959 | +1.2 | 1 of 1 | 0.9 |
| Nielsen | 1 | 1992 | +7.4 | 1 of 1 | 2.8 |
| NMR | 1 | 1992 | +8.6 | 1 of 1 | 3.1 |
| Rasmussen | 1 | 2001 | +3.6 | 1 of 1 | 2.1 |
| Angus Reid | 1 | 2010 | −4.8 | 0 of 1 | 4.0 |
| OnePoll | 1 | 2010 | −1.8 | 0 of 1 | 8.0 |
| RNB | 1 | 2010 | −1.8 | 0 of 1 | 1.4 |
| ORB | 1 | 2017 | −6.6 | 0 of 1 | 2.3 |
| Focaldata | 1 | 2024 | +11.7 | 1 of 1 | 4.8 |
| JL Partners | 1 | 2024 | +4.7 | 1 of 1 | 1.7 |
| More in Common | 1 | 2024 | +4.7 | 1 of 1 | 1.8 |
| PeoplePolling | 1 | 2024 | +9.7 | 1 of 1 | 4.1 |
| Redfield & Wilton | 1 | 2024 | +8.7 | 1 of 1 | 3.8 |
| Stonehaven | 1 | 2024 | +7.3 | 1 of 1 | 3.1 |
| Techne | 1 | 2024 | +11.7 | 1 of 1 | 4.1 |
| WeThink | 1 | 2024 | +7.7 | 1 of 1 | 3.1 |
| Whitestone | 1 | 2024 | +6.7 | 1 of 1 | 3.1 |
Every final poll
| Election | Company | Fieldwork | Con | Lab | Lib/LD | Green | Reform/UKIP | Lead error | Notes |
|---|---|---|---|---|---|---|---|---|---|
| 1945 | Gallup | 24-27 Jun | 41 +1.7 | 47 −1.8 | 10.5 | −3.5 | for News Chron | ||
| 1950 | Daily Mail (own poll) | 5-11 Feb | 45.5 +2.5 | 42.5 −4.3 | 12 | −6.8 | as Daily Mail | ||
| 1950 | Daily Express (own poll) | 17-21 Feb | 44.5 +1.5 | 44 −2.8 | 11 | −4.3 | newspaper's own poll; as DE; for DE | ||
| 1950 | Gallup | 17-20 Feb | 43.5 +0.5 | 45 −1.8 | 10.5 | −2.3 | for News Chron | ||
| 1951 | Daily Express (own poll) | 19-23 Oct | 50 +2.2 | 46 −3.4 | 3.5 | −5.6 | newspaper's own poll; as DE; for DE | ||
| 1951 | Research Services | 19 Oct | 50 +2.2 | 43 −6.4 | 6.7 | −8.6 | as Research Serv Ltd; for DG | ||
| 1951 | Gallup | 22 Oct | 49.5 +1.7 | 47 −2.4 | 3 | −4.1 | for News Chron | ||
| 1955 | Gallup | 21-24 May | 51 +1.7 | 47.5 +0.1 | 1.5 | −1.6 | for News Chron | ||
| 1959 | Forecasting Statistics | 1-4 Oct | 49 +0.2 | 46 +1.4 | 5 | +1.2 | for DT | ||
| 1959 | NOP | 2-5 Oct | 48 −0.8 | 44.1 −0.5 | 7.9 | +0.3 | for DM | ||
| 1959 | Gallup | 3-6 Oct | 48.5 −0.3 | 46.5 +1.9 | 4.5 | +2.2 | for News Chron | ||
| 1959 | Daily Express (own poll) | 7 Oct | 49.1 +0.3 | 45.4 +0.8 | 5 | +0.5 | newspaper's own poll; as DE; for DE | ||
| 1964 | Research Services | 1 Oct | 46 +3.1 | 47 +2.2 | 7 | −0.9 | as Research Serv Ltd | ||
| 1964 | NOP | 9-12 Oct | 44.3 +1.4 | 47.4 +2.6 | 7.9 | +1.2 | for DM | ||
| 1964 | Daily Express (own poll) | 10 Oct | 44.5 +1.6 | 43.7 −1.1 | 11.1 | −2.7 | newspaper's own poll; as DE; for DE | ||
| 1964 | Gallup | 13 Oct | 44.5 +1.6 | 46 +1.2 | 9 | −0.4 | |||
| 1966 | Research Services | 22 Mar | 41.6 +0.2 | 49.7 +0.9 | 8.3 | +0.7 | as Research Serv Ltd; for Ob | ||
| 1966 | Gallup | 24-28 Mar | 40 −1.4 | 51 +2.2 | 8 | +3.6 | for DT | ||
| 1966 | Daily Express (own poll) | 27 Mar | 37.4 −4.0 | 54.1 +5.3 | 7.7 | +9.3 | newspaper's own poll; as DE; for DE | ||
| 1966 | NOP | 27-29 Mar | 41.6 +0.2 | 50.6 +1.8 | 7.4 | +1.6 | for DM | ||
| 1970 | Marplan | 11-14 Jun | 41.5 −4.7 | 50.2 +6.3 | 7 | +11.0 | for TT | ||
| 1970 | NOP | 12-16 Jun | 44.1 −2.1 | 48.2 +4.3 | 6.4 | +6.4 | for DM | ||
| 1970 | ORC | 13-17 Jun | 46.5 +0.3 | 45.5 +1.6 | 6.5 | +1.3 | for ES | ||
| 1970 | Gallup | 14-16 Jun | 42 −4.2 | 49 +5.1 | 7.5 | +9.3 | for DT | ||
| 1970 | Harris | 14-16? Jun | 46 −0.2 | 48 +4.1 | 5 | +4.3 | for DE | ||
| 1974 Feb | Business Decisions | 21 Feb | 35 −3.8 | 37.5 −0.5 | 23.5 | +3.3 | for Ob | ||
| 1974 Feb | NOP | 23-27 Feb | 39.5 +0.7 | 35.5 −2.5 | 22 | −3.2 | for DM | ||
| 1974 Feb | Gallup | 26-27 Feb | 39.5 +0.7 | 37.5 −0.5 | 20.5 | −1.2 | for DT | ||
| 1974 Feb | Harris | 26-27 Feb | 40.2 +1.4 | 35.2 −2.8 | 22 | −4.2 | for DE | ||
| 1974 Feb | Marplan | 27 Feb | 36.5 −2.3 | 34.5 −3.5 | 25 | −1.2 | for BEM / LWT | ||
| 1974 Feb | ORC | 27 Feb | 38.7 −0.1 | 34.9 −3.1 | 23 | −3.0 | for TT | ||
| 1974 Oct | Business Decisions | 2 Oct | 35.5 −1.2 | 40 −0.2 | 20 | +1.0 | for Ob | ||
| 1974 Oct | NOP | 2-5 Oct | 31 −5.7 | 45.5 +5.3 | 19.5 | +11.0 | for DM | ||
| 1974 Oct | Gallup | 3-7 Oct | 36 −0.7 | 41.5 +1.3 | 19 | +2.0 | for DT | ||
| 1974 Oct | Harris | 8-9 Oct | 34.6 −2.1 | 43 +2.8 | 19.3 | +4.9 | for DE | ||
| 1974 Oct | Marplan | 8 Oct | 33.3 −3.4 | 43.8 +3.6 | 19.5 | +7.0 | for Sun | ||
| 1974 Oct | ORC | 9 Oct | 34.4 −2.3 | 41.8 +1.6 | 19.4 | +3.9 | for ES | ||
| 1979 | Research Services | 24-25 Apr | 49.5 +4.6 | 38 +0.2 | 10 | −4.4 | as Research Srv; for Ob | ||
| 1979 | Marplan | 1 May | 45 +0.1 | 38.5 +0.7 | 13.5 | +0.6 | for Sun | ||
| 1979 | NOP | 1-2 May | 46 +1.1 | 39 +1.2 | 12.5 | +0.1 | for DM | ||
| 1979 | Gallup | 2-3 May | 45.6 +0.7 | 37.9 +0.1 | 13.5 | −0.6 | for BBC | ||
| 1979 | Ipsos | 2 May | 45 +0.1 | 37 −0.8 | 15 | −0.9 | as MORI; for ES | ||
| 1979 | ORC | 3 May | 44 −0.9 | 38 +0.2 | 14 | +1.1 | for ITN | ||
| 1983 | ASL | 7 Jun | 46 +2.5 | 23 −5.3 | 29 | −7.8 | for Sun | ||
| 1983 | Gallup | 7-8 Jun | 45.5 +2.0 | 26.5 −1.8 | 26 | −3.8 | for DT | ||
| 1983 | Harris | 7-8 Jun | 47 +3.5 | 25 −3.3 | 26 | −6.8 | for Ob | ||
| 1983 | Ipsos | 8 Jun | 44 +0.5 | 28 −0.3 | 26 | −0.8 | as MORI; for ES | ||
| 1983 | Marplan | 8 Jun | 46 +2.5 | 26 −2.3 | 26 | −4.8 | for Gu | ||
| 1983 | NOP | 8 Jun | 47 +3.5 | 25 −3.3 | 26 | −6.8 | for Northcliffe | ||
| 1987 | ASL | 8 Jun | 43 −0.3 | 34 +2.5 | 21 | +2.8 | for Sun | ||
| 1987 | Gallup | 8-9 Jun | 41 −2.3 | 34 +2.5 | 23.5 | +4.8 | for DT | ||
| 1987 | Harris | 8-9 Jun | 42 −1.3 | 35 +3.5 | 21 | +4.8 | for TV-AM | ||
| 1987 | Ipsos | 9-10 Jun | 44 +0.7 | 32 +0.5 | 22 | −0.2 | as MORI; for TT | ||
| 1987 | Marplan | 9 Jun | 42 −1.3 | 35 +3.5 | 21 | +4.8 | for Gu | ||
| 1987 | NOP | 10 Jun | 42 −1.3 | 35 +3.5 | 21 | +4.8 | for Ind | ||
| 1992 | NMR | 24 Mar | 38 −4.8 | 39 +3.8 | 19 | +8.6 | for TE | ||
| 1992 | Nielsen | 28 Mar | 39.8 −3.0 | 39.6 +4.4 | 17.2 | +7.4 | as NEILSEN | ||
| 1992 | ASL | 3-4 Apr | 35 −7.8 | 38 +2.8 | 23 | +10.6 | |||
| 1992 | Harris | 4-7 Apr | 38 −4.8 | 40 +4.8 | 18 | +9.6 | for ITN | ||
| 1992 | Gallup | 7-8 Apr | 38.5 −4.3 | 38 +2.8 | 20 | +7.1 | for DT | ||
| 1992 | Ipsos | 7-8 Apr | 38 −4.8 | 39 +3.8 | 20 | +8.6 | as MORI; for TT | ||
| 1992 | NOP | 7-8 Apr | 39 −3.8 | 42 +6.8 | 17 | +10.6 | for Ind | ||
| 1992 | ICM | 8 Apr | 38 −4.8 | 38 +2.8 | 20 | +7.6 | for GU | ||
| 1997 | Harris | 28-29 Apr | 31 −0.5 | 48 +3.7 | 15 | +4.2 | for Ind | ||
| 1997 | Gallup | 29-30 Apr | 33 +1.5 | 46 +1.7 | 16 | +0.2 | for DT | ||
| 1997 | ICM | 29-30 Apr | 33 +1.5 | 43 −1.3 | 18 | −2.8 | for Gu | ||
| 1997 | NOP | 29 Apr | 28 −3.5 | 50 +5.7 | 14 | +9.2 | for Reut | ||
| 1997 | Ipsos | 30 Apr | 29 −2.5 | 46 +1.7 | 19 | +4.2 | as MORI; for ES | ||
| 2001 | NOP | 31-1 May | 30 −2.6 | 47 +5.0 | 16 | +7.6 | for ST | ||
| 2001 | ICM | 2-4 Jun | 32 −0.6 | 43 +1.0 | 19 | +1.6 | for Gu | ||
| 2001 | Rasmussen | 2-3 Jun | 32 −0.6 | 45 +3.0 | 16 | +3.6 | for Ind | ||
| 2001 | YouGov | 4 Jun | 33 +0.4 | 43 +1.0 | 17 | +0.6 | |||
| 2001 | Ipsos | 5-6 Jun | 30 −2.6 | 45 +3.0 | 18 | +5.6 | as MORI; for TT | ||
| 2001 | Gallup | 6 Jun | 32 −0.6 | 48 +6.0 | 18 | +6.6 | for DT | ||
| 2005 | ComRes / Savanta | 23-28 Apr | 31 −2.2 | 39 +2.9 | 23 | +5.1 | as CommR; for IoS | ||
| 2005 | BPIX | 28-30 Apr | 31 −2.2 | 38 +1.9 | 22 | +4.1 | for MoS | ||
| 2005 | Harris | 1-4 May | 33 −0.2 | 39 +2.9 | 22 | +3.1 | |||
| 2005 | ICM | 1-3 May | 32 −1.2 | 38 +1.9 | 22 | +3.1 | for Gu | ||
| 2005 | NOP | 1-3 May | 33 −0.2 | 36 −0.1 | 23 | +0.1 | for Ind | ||
| 2005 | Populus | 2-3 May | 32 −1.2 | 38 +1.9 | 21 | +3.1 | for TT | ||
| 2005 | Ipsos | 3-4 May | 33 −0.2 | 38 +1.9 | 23 | +2.1 | as MORI; for ES | ||
| 2005 | YouGov | 3-4 May | 32 −1.2 | 37 +0.9 | 24 | +2.1 | for DT | ||
| 2010 | Kantar / Verian | 29-4 Apr | 33 −3.9 | 27 −2.7 | 29 | +1.2 | as TNS-BMRB | ||
| 2010 | OnePoll | 29-1 ? Apr | 30 −6.9 | 21 −8.7 | 32 | −1.8 | for S People | ||
| 2010 | BPIX | 30-1 Apr | 34 −2.9 | 27 −2.7 | 30 | +0.2 | for MoS | ||
| 2010 | RNB | 30-2 Apr | 37 +0.1 | 28 −1.7 | 26 | −1.8 | |||
| 2010 | ICM | 3-4 May | 36 −0.9 | 28 −1.7 | 26 | −0.8 | for Gu | ||
| 2010 | Angus Reid | 4-5 May | 36 −0.9 | 24 −5.7 | 29 | −4.8 | as Angus RS; for PB | ||
| 2010 | ComRes / Savanta | 4-5 May | 37 +0.1 | 28 −1.7 | 28 | −1.8 | as ComRes; for ITV/Ind | ||
| 2010 | Harris | 4-5 May | 35 −1.9 | 29 −0.7 | 27 | +1.2 | for DM | ||
| 2010 | Opinium | 4-5 May | 35 −1.9 | 27 −2.7 | 26 | −0.8 | for DE | ||
| 2010 | Populus | 4-5 May | 37 +0.1 | 28 −1.7 | 27 | −1.8 | for TT | ||
| 2010 | YouGov | 4-5 May | 35 −1.9 | 28 −1.7 | 28 | +0.2 | for Sun | ||
| 2010 | Ipsos | 5 May | 36 −0.9 | 29 −0.7 | 27 | +0.2 | as MORI; for ES | ||
| 2015 | Kantar / Verian | 30-4 Apr | 33 −4.7 | 32 +0.8 | 8 | 6 | 14 | +5.5 | as TNS BMRB |
| 2015 | SurveyMonkey | 30-6 Apr | 34 −3.7 | 28 −3.2 | 7 | 8 | 13 | +0.5 | for Wash Po |
| 2015 | Panelbase | 1-6 May | 31 −6.7 | 33 +1.8 | 8 | 5 | 16 | +8.5 | |
| 2015 | BMG | 3-5 May | 34 −3.7 | 34 +2.8 | 10 | 4 | 12 | +6.5 | for May2015 |
| 2015 | ICM | 3-6 May | 34 −3.7 | 35 +3.8 | 9 | 4 | 11 | +7.5 | for Gu |
| 2015 | Opinium | 4-5 May | 35 −2.7 | 34 +2.8 | 8 | 6 | 12 | +5.5 | |
| 2015 | YouGov | 4-6 May | 34 −3.7 | 34 +2.8 | 10 | 4 | 12 | +6.5 | |
| 2015 | Ipsos | 5-6 May | 36 −1.7 | 35 +3.8 | 8 | 5 | 11 | +5.5 | as MORI; for ES |
| 2015 | Lord Ashcroft | 5-6 May | 33 −4.7 | 33 +1.8 | 10 | 6 | 11 | +6.5 | |
| 2015 | Populus | 5-7 May | 33 −4.7 | 33 +1.8 | 10 | 5 | 14 | +6.5 | |
| 2015 | ComRes / Savanta | 6 May | 35 −2.7 | 34 +2.8 | 9 | 4 | 12 | +5.5 | as ComRes; for DM/ITV |
| 2015 | Survation | 6 May | 37 −0.7 | 31 −0.2 | 10 | 5 | 11 | +0.5 | |
| 2017 | Norstat | 31-1 May | 39 −4.4 | 35 −6.0 | 8 | 3 | 6 | −1.6 | for SE |
| 2017 | ORB | 31-1 May | 45 +1.6 | 36 −5.0 | 8 | 4 | −6.6 | for STel | |
| 2017 | Kantar / Verian | 1-7 Jun | 43 −0.4 | 38 −3.0 | 7 | 2 | 4 | −2.6 | as Kantar TNS |
| 2017 | Panelbase | 2-7 Jun | 44 +0.6 | 36 −5.0 | 7 | 2 | 5 | −5.6 | |
| 2017 | Opinium | 4-6 Jun | 43 −0.4 | 36 −5.0 | 8 | 2 | 5 | −4.6 | |
| 2017 | Qriously | 4-7 Jun | 39 −4.4 | 41 0.0 | 8 | 1 | 4 | +4.4 | |
| 2017 | SurveyMonkey | 4-6 Jun | 42 −1.4 | 38 −3.0 | 6 | 2 | 4 | −1.6 | for Sun |
| 2017 | ComRes / Savanta | 5-7 Jun | 44 +0.6 | 34 −7.0 | 9 | 2 | 5 | −7.6 | as ComRes; for Ind |
| 2017 | YouGov | 5-7 Jun | 42 −1.4 | 35 −6.0 | 10 | 2 | 5 | −4.6 | for TT |
| 2017 | BMG | 6-7 Jun | 46 +2.6 | 33 −8.0 | 8 | 3 | 5 | −10.6 | for Herald |
| 2017 | ICM | 6-7 Jun | 46 +2.6 | 34 −7.0 | 7 | 2 | 5 | −9.6 | for Gu |
| 2017 | Ipsos | 6-7 Jun | 44 +0.6 | 36 −5.0 | 7 | 2 | 4 | −5.6 | as Ipsos MORI; for ES |
| 2017 | Survation | 6-7 Jun | 41 −2.4 | 40 −1.0 | 8 | 2 | 2 | +1.4 | |
| 2019 | Qriously | 5-8 Dec | 43 −1.7 | 30 −2.9 | 12 | 4 | 3 | −1.2 | |
| 2019 | YouGov | 5-6 Dec | 43 −1.7 | 33 +0.1 | 13 | 3 | 3 | +1.8 | for ST |
| 2019 | BMG | 6-11 Dec | 41 −3.7 | 32 −0.9 | 14 | 4 | 3 | +2.8 | for Ind |
| 2019 | ICM | 6-9 Dec | 42 −2.7 | 36 +3.1 | 12 | 2 | 3 | +5.8 | for Reut |
| 2019 | Number Cruncher | 8-10 Dec | 43 −1.7 | 33 +0.1 | 12 | 3 | 3 | +1.8 | as NCP; for Bloom |
| 2019 | ComRes / Savanta | 9-10 Dec | 41 −3.7 | 36 +3.1 | 12 | 2 | 3 | +6.8 | as ComRes; for DT |
| 2019 | Deltapoll | 9-11 Dec | 45 +0.3 | 35 +2.1 | 10 | 3 | 3 | +1.8 | |
| 2019 | Ipsos | 9-11 Dec | 44 −0.7 | 33 +0.1 | 12 | 3 | 2 | +0.8 | as Ipsos MORI; for ES |
| 2019 | Kantar / Verian | 9-11 Dec | 44 −0.7 | 32 −0.9 | 13 | 3 | 3 | −0.2 | as Kantar TNS |
| 2019 | Opinium | 10-11 Dec | 45 +0.3 | 33 +0.1 | 12 | 2 | 2 | −0.2 | |
| 2019 | Panelbase | 10-11 Dec | 43 −1.7 | 34 +1.1 | 11 | 4 | 3 | +2.8 | |
| 2019 | Survation | 10-11 Dec | 44.5 −0.2 | 33.7 +0.8 | 9.3 | 3.2 | 3.1 | +1.0 | |
| 2024 | Focaldata | 14-17 Jun | 21 −3.4 | 43 +8.3 | 10 | 5 | 16 | +11.7 | |
| 2024 | YouGov | 25-27 Jun | 20 −4.4 | 37 +2.3 | 13 | 7 | 17 | +6.7 | |
| 2024 | Techne | 26-27 Jun | 19 −5.4 | 41 +6.3 | 12 | 5 | 17 | +11.7 | for Ind |
| 2024 | Lord Ashcroft | 27-1 Jun | 19 −5.4 | 38 +3.3 | 11 | 8 | 18 | +8.7 | for DM |
| 2024 | Stonehaven | 27-30 Jun | 21.9 −2.5 | 39.5 +4.8 | 10.5 | 6.5 | 15.5 | +7.3 | |
| 2024 | Kantar / Verian | 28-1 Jun | 21 −3.4 | 36 +1.3 | 13 | 7 | 16 | +4.7 | as Verian |
| 2024 | More in Common | 28-30 Jun | 24 −0.4 | 39 +4.3 | 12 | 5 | 15 | +4.7 | |
| 2024 | Redfield & Wilton | 28-2 Jun | 22 −2.4 | 41 +6.3 | 10 | 6 | 16 | +8.7 | as R+W |
| 2024 | Deltapoll | 29-3 Jun | 22 −2.4 | 39 +4.3 | 10 | 7 | 17 | +6.7 | |
| 2024 | BMG | 30-2 Jun | 22 −2.4 | 39 +4.3 | 11 | 7 | 16 | +6.7 | for The i |
| 2024 | Ipsos | 1-3 Jul | 19 −5.4 | 37 +2.3 | 11 | 9 | 15 | +7.7 | for ES |
| 2024 | Norstat | 1-3 Jul | 24 −0.4 | 37 +2.3 | 11 | 6 | 16 | +2.7 | |
| 2024 | Opinium | 1-3 Jul | 21 −3.4 | 41 +6.3 | 11 | 7 | 17 | +9.7 | |
| 2024 | Survation | 1-3 Jul | 20 −4.4 | 38 +3.3 | 12 | 7 | 17 | +7.7 | |
| 2024 | Whitestone | 1-2 Jul | 21 −3.4 | 38 +3.3 | 10 | 7 | 18 | +6.7 | for DE/Mirror |
| 2024 | ComRes / Savanta | 2-3 Jul | 20 −4.4 | 39 +4.3 | 10 | 5 | 17 | +8.7 | as Savanta ComRes; for DT |
| 2024 | JL Partners | 2-3 Jul | 23 −1.4 | 38 +3.3 | 13 | 5 | 17 | +4.7 | for TRIP |
| 2024 | PeoplePolling | 2 Jul | 16 −8.4 | 36 +1.3 | 10 | 9 | 20 | +9.7 | for GB News |
| 2024 | WeThink | 2-3 Jul | 23 −1.4 | 41 +6.3 | 11 | 7 | 15 | +7.7 | |
| 2024 | Number Cruncher | 3 Jul | 23 −1.4 | 41 +6.3 | 11 | 7 | 13 | +7.7 | as NCP |
How good is the old data?
- Britain only. No England, Scotland or Wales split, so it can't be compared with our England record directly or used in the correction.
- Headline figures only. There are no raw tables, so we can't test starting from the raw answers before 2015.
- Different methods. Face-to-face quota samples until the late 1980s, then telephone, then online panels from the 2000s. An old miss tells you how polling has failed, not how today's companies will.
- Few polls early on. Gallup alone in 1945 and 1955; three to six companies until 1992; twelve or more since 2010.
- Rounding. Most shares were published as whole numbers until the 2000s, which adds up to half a point either way.
- Which poll counts as final. We take each company's last poll with fieldwork starting in the three weeks before polling day. Exit polls, MRP seat models and polls released after the election are left out. Renamed companies are merged (MORI / Ipsos MORI / Ipsos, ComRes / Savanta, TNS BMRB / Kantar / Verian). Early polls run by newspapers themselves are marked.
Source: PollBase by Mark Pack (Q2 2026 edition), used with credit as he asks. GB results from PollBase's result rows.
Final GB polls by pollster (England figures)
| Election | Pollster | Fieldwork end | England figure | In benchmark | Note |
|---|---|---|---|---|---|
| 2024 | BMG Research | 2 Jul | Unavailable | No | Data tables unrecoverable: linked XLSX (bmgresearch.co.uk/wp-content/uploads/2024/07/GE-Week-6-Polling-for-the-i-Newspaper-v1.xlsx) now 302-redirects to the BMG news index on both .co.uk and .com domains; web.archive.org is blocked for WebFetch. |
| 2024 | Deltapoll | 3 Jul | Reconstructed | No | |
| 2024 | Find Out Now (with Electoral Calculus) | 24 Jun | Unavailable | No | Only an MRP (with Electoral Calculus) in the final period; no standard VI poll |
| 2024 | Focaldata | 1 Jul | Unavailable | No | MRP-derived final estimate, not a standard voting-intention poll |
| 2024 | Ipsos | 3 Jul | Direct | Yes | |
| 2024 | J.L. Partners | 3 Jul | Reconstructed | Yes | |
| 2024 | Lord Ashcroft Polls | 1 Jul | Unavailable | No | No data-table link found on the final-poll posts or the 2024 GE category page; tables not locatable by search; archive.org blocked. |
| 2024 | More in Common | 30 Jun | Unavailable | No | The 'June 2024 polling tables' URL now serves the general archive (Dec 2024 onward); no June/July 2024 table links found. More in Common tables are XLSX, which WebFetch cannot parse. |
| 2024 | Norstat | 3 Jul | Reconstructed | Yes | |
| 2024 | Number Cruncher Politics | 3 Jul | Reconstructed | No | |
| 2024 | Opinium | 3 Jul | Reconstructed | Yes | |
| 2024 | PeoplePolling | 2 Jul | Unavailable | No | No public data tables located (Matt Goodwin poll archive page returns 404; no PeoplePolling tables found by search). |
| 2024 | Prolific | 2 Jul | Unavailable | No | No data tables located (prolific.com/polling gives no 2 July 2024 tables); poll not listed in Wikipedia table. |
| 2024 | Redfield & Wilton Strategies | 2 Jul | Unavailable | No | Article URL now redirects to R&W homepage; historical tables moved behind paid 'Research Archive'; web.archive.org is blocked for WebFetch. Tables unrecoverable. |
| 2024 | Savanta | 3 Jul | Direct | Yes | |
| 2024 | Stonehaven | 30 Jun | Unavailable | No | Final survey is labelled as the MRP poll; published crosstab has region columns (NE, NW, EM, EoE, GL, SE, SW, Wales, Scotland, Yorks, WM) but NO weighted bases, includes 'I wouldn't vote' in percentages, and its 'All' column (Con 20, Lab 36, LD 10, Grn 6, Ref 14) does not match the published headline (BPC: 24/39/12/5/15). Cannot reconstruct without assumptions. |
| 2024 | Survation | 3 Jul | Unavailable | No | Tables host files.survation.com is disallowed by robots.txt for WebFetch; archive.org blocked. BPC post-mortem (ge24-survation-cati-postmortem.pdf) gives no England VI. |
| 2024 | Techne UK | 2 Jul | Unavailable | No | Techne archive and wp-content uploads return HTTP 401 to WebFetch; archive.org blocked; data file URLs not discoverable. |
| 2024 | Verian | 1 Jul | Unavailable | No | Tables obtained; headline VI table (Total matches headline) has only gender, age, 2019 vote and EU ref crossbreaks - no region/country columns in any table. England cannot be isolated. |
| 2024 | We Think (Omnisis) | 3 Jul | Unavailable | No | The workbook ('TRACKING VI POLL', 'Prepared by European Movement : 03 July 24') contains ONLY question WK5Q2 - approval of Rishi Sunak as PM - by week, with crossbreak sheets for gender, age, Con/Lab/LD VI, 11 GB regions, Red Wall and party-target seats. It contains no voting-intention question at all, so the headline cannot be located or reproduced and no England VI can be derived. The file does confirm the final wave (column ' 3rd July 2024', unweighted 1325 / weighted 1308) and that regional sheets have weighted bases (9 English regions + Wales + Scotland, no NI), so England WOULD be reconstructable if the VI tables for that wave were obtained. Confidence 'High' refers to the Unavailable determination. The file appears to be the wrong download (approval tracker rather than VI tables); the VI tables (e.g. 240703_Voter_Intention xlsx) should be re-sourced. |
| 2024 | Whitestone Insight | 2 Jul | Unavailable | No | Published 'final tables' document is a single-page topline trend table with no crossbreaks (no regional columns). |
| 2024 | YouGov | 27 Jun | Reconstructed | Yes | |
| 2019 | BMG Research | 11 Dec | Unavailable | No | Old bmgresearch.co.uk poll pages redirect to bmgresearch.com/news; tables not locatable; Wayback blocked. |
| 2019 | Deltapoll | 11 Dec | Reconstructed | No | |
| 2019 | Focaldata | 10 Dec | Unavailable | No | MRP model output, not a standard voting-intention poll |
| 2019 | ICM Research | 9 Dec | Unavailable | No | icmunlimited.com now redirects to walnutunlimited.com; tables unrecoverable; Wayback blocked. |
| 2019 | Ipsos MORI | 11 Dec | Direct | Yes | |
| 2019 | Number Cruncher Politics | 10 Dec | Reconstructed | No | |
| 2019 | Opinium | 11 Dec | Reconstructed | Yes | |
| 2019 | Panelbase | 11 Dec | Unavailable | No | Release page and all drg.global/panelbase PDF URLs returned empty content via WebFetch (panelbase.com robots fetch timed out); tables could not be retrieved. |
| 2019 | Qriously | 8 Dec | Reconstructed | Yes | |
| 2019 | Savanta ComRes | 10 Dec | Direct | Yes | |
| 2019 | Survation | 11 Dec | Direct | Yes | |
| 2019 | Kantar | 11 Dec | Unavailable | No | Release page links to 'survey data' but the link target is not present in the fetched page; no tables found by search and the Wayback Machine is blocked. |
| 2019 | YouGov | 6 Dec | Unavailable | No | England cannot be isolated: region banner combines Midlands and Wales; no England or Wales column. |
| 2017 | BMG Research | 7 Jun | Unavailable | No | Tables unrecoverable: bmgresearch.co.uk article redirects to bmgresearch.com/news (original post 404); wp-content/uploads/2017 directories 404; archive.org blocked. |
| 2017 | ICM Unlimited | 7 Jun | Unavailable | No | Tables unrecoverable: icmunlimited.com now redirects every URL to walnutunlimited.com; web.archive.org and archive.ph are blocked for this tool; no alternative official copy found. |
| 2017 | Ipsos MORI | 7 Jun | Direct | Yes | |
| 2017 | Opinium | 6 Jun | Direct | Yes | |
| 2017 | Panelbase | 7 Jun | Unavailable | No | Tables not located: panelbase.com unreachable (timeout); no copy found via search; archive.org blocked. |
| 2017 | Qriously | 7 Jun | Unavailable | No | No detailed tables located (no Qriously/Wired table document found via search). |
| 2017 | ComRes | 7 Jun | Direct | Yes | |
| 2017 | Survation | 7 Jun | Direct | Yes | |
| 2017 | SurveyMonkey | 6 Jun | Unavailable | No | No detailed tables located; SurveyMonkey election blog (blog.electiontracking.surveymonkey.com) fails TLS; not a BPC member so no BPC-required tables. |
| 2017 | Kantar Public | 7 Jun | Unavailable | No | Tables unrecoverable: tables URL on tns-bmrb.co.uk (domain dead, connection timeout); uk.kantar.com press page redirects to 404; archive.org blocked. |
| 2017 | YouGov | 7 Jun | Unavailable | No | Region banner combines Midlands and Wales ('Midlands/Wales'); Wales cannot be separated and there is no England column, so England cannot be isolated without assumptions. |
| 2015 | BMG Research | 5 May | Direct | No | |
| 2015 | ICM | 6 May | Direct | No | |
| 2015 | Ipsos MORI | 6 May | Unavailable | No | Final 2015 computer tables not locatable on ipsos.com (only the 2010 final-poll page/tables are indexed; political monitor archive has no April/May 2015 VI entry); Wayback Machine blocked. |
| 2015 | Lord Ashcroft Polls | 6 May | Direct | No | |
| 2015 | Opinium | 5 May | Direct | No | |
| 2015 | Panelbase | 6 May | Unavailable | No | No data tables for this poll could be located on the web. |
| 2015 | Populus | 7 May | Unavailable | No | England cannot be isolated: region banner is South East / Midlands / North England / Wales & South West / Scotland, so Wales is merged with the South West. |
| 2015 | ComRes | 6 May | Direct | No | |
| 2015 | Survation | 6 May | Direct | Yes | |
| 2015 | SurveyMonkey | 6 May | Unavailable | No | No crossbreak tables located; poll was reported only as toplines (Washington Post). |
| 2015 | TNS UK (TNS BMRB; later Kantar Public) | 4 May | Unavailable | No | No data tables for this poll could be located on the web. |
| 2015 | YouGov | 6 May | Direct | Yes |
Figures as of . Published by Polling Forecast.





