Your polling error
Points by which the polls are wrong about each bloc's round-one share. Shares are rescaled to the same total.
Seats that change hands
How big is a realistic polling error?
The biggest miss here: the polls had Ensemble and centre 1.8 points too low (2017). On average they have put Reconquête 1.1 points too high, and our forecast allows for that.
| Party | 2002 5 polls | 2007 5 polls | 2012 9 polls | 2017 8 polls | 2022 7 polls | 2024 7 polls | Average | We adjust |
|---|---|---|---|---|---|---|---|---|
| Ensemble and centre | – | – | – | −1.8 | +1.0 | −0.6 | −0.5 | +0.2 |
| Reconquête | – | – | – | – | +1.6 | +0.7 | +1.1 | −0.6 |
| Far left | +0.7 | 0.0 | +0.4 | +0.1 | 0.0 | −0.1 | +0.2 | −0.1 |
Average of each pollster's final poll minus the result, in points. + means the polls overstated the party, − that they understated it.
If the polls miss like a past election
Today's polls in France with each past election's polling error applied, through our seat model. Our forecast: Left (NFP and other left) 291 of 577, a majority of 5.
| If the polls miss like | Largest party | Seats | Majority? | Seats change hands |
|---|---|---|---|---|
| 2002 | Left (NFP and other left) | 278 | 11 short of a majority | 27 |
| 2007 | Left (NFP and other left) | 291 | a majority of 5 | 7 |
| 2012 | Left (NFP and other left) | 319 | a majority of 61 | 33 |
| 2017 | Left (NFP and other left) | 318 | a majority of 59 | 50 |
| 2022 | Left (NFP and other left) | 287 | 2 short of a majority | 24 |
| 2024 | Left (NFP and other left) | 300 | a majority of 23 | 28 |
Figures as of . Published by Polling Forecast.




