Presidential Election 2027Probabilities of winning, recalculated every day
FRRUEN

Rejection of candidates in polls

Experimental

Experimental. This page is experimental and outside P(win): none of these figures enters the model. It shows what two pollsters published about the rejection of each public figure, nothing more.

Why it is outside the model

We could not test this measure on 2017 and 2022: those series have not been collected, and two elections (N ≈ 2) would not be enough to show it adds anything anyway. Until it is tested, it stays here, on its own.

What the page shows

Two series, each with its own question. They do not measure the same thing, so they are neither merged nor compared with each other. For Elabe, the two negative answers are given separately, as the pollster publishes them; we do not add them up.

Other pollsters publish nearby measures and are not shown: reusing Ifop's figures needs its express agreement, and Toluna Harris Interactive measures something else (a voting potential, not a rejection).

Odoxa: "You reject her/him"

Experimental

Pollster
Odoxa
Client and frame
Mascaret, Public Sénat et presse régionale
Question asked (the pollster's text)
« Pour chacune des personnalités politiques suivantes, dites-nous si vous la soutenez, si vous éprouvez de la sympathie pour elle, si vous ressentez de l’indifférence à son égard ou si vous la rejetez. »
Waves

Margin of error: about ±3.1 points for the latest sample (n = 1005), roughly, without allowing for quotas. A gap of one or two points between two waves can be sample noise alone. The reports point to each poll's technical notice at the Commission des sondages. Commission des sondages.

Share of French people who reject the figure, wave by wave

Odoxa: "You reject her/him" Rejection lines for 8 candidates, from 22 Jan 2026 to 22 Oct 2026. 304050607080 1 Feb1 Apr1 Jun1 Aug1 Oct Mélenchon 68% Tondelier 53% Le Pen 46% Bardella 46% Attal 43% Retailleau 42% Philippe 41% Glucksmann 38%

The chart shows the candidates who have a colour elsewhere on the site, named at the end of their line; the table below gives all of them. The axis does not start at zero. Where a wave is missing, the line simply joins the two neighbouring waves.

All candidates, wave by wave (% who reject)
Candidate22 Jan19 Feb26 Mar23 Apr21 May25 Jun24 Sep22 Oct
Gabriel Attal40%40%36%39%42%41%45%43%
Jordan Bardella45%47%43%49%48%46%48%46%
Xavier Bertrand39%40%—42%42%43%——
Bernard Cazeneuve————40%39%——
Olivier Faure43%43%41%45%45%44%47%44%
Raphaël Glucksmann40%41%34%43%42%41%43%38%
François Hollande45%—43%45%46%44%47%46%
Sarah Knafo———50%50%47%52%48%
Marine Le Pen47%49%47%49%53%48%49%46%
David Lisnard38%36%30%38%43%—40%36%
Jean-Luc Mélenchon69%71%68%72%69%69%69%68%
Édouard Philippe38%39%33%40%39%39%45%41%
Bruno Retailleau41%41%38%44%43%42%46%42%
Fabien Roussel39%41%35%42%41%41%46%42%
François Ruffin40%40%36%41%41%42%44%42%
Marine Tondelier49%48%45%50%48%53%52%53%
Dominique de Villepin40%——41%45%44%46%42%
Laurent Wauquiez48%47%43%48%49%48%——
Éric Zemmour———66%67%65%68%64%

Elabe: negative image

Experimental

Pollster
Elabe
Client and frame
Les Echos, Observatoire politique
Question asked (the pollster's text)
« Pour chacune des personnalités suivantes, indiquez si vous en avez une image positive ou une image négative ? »
Waves

Margin of error: about ±3.1 points for the latest sample (n = 1005), roughly, without allowing for quotas. A gap of one or two points between two waves can be sample noise alone. The reports point to each poll's technical notice at the Commission des sondages. Commission des sondages.

Latest wave (2 Sep 2026): the two answers, side by side

  • Very negative image
  • Rather negative image
Elabe: negative image 19 bars, one per candidate, with the share of very negative then rather negative image. Attal 32% 27% Bardella 41% 11% Bertrand 29% 23% Cazeneuve 27% 23% Faure 28% 24% Glucksmann 28% 21% Hollande 37% 27% Knafo 32% 12% Le Pen 42% 10% Lisnard 22% 13% Mélenchon 60% 12% Philippe 31% 26% Retailleau 32% 19% Roussel 25% 22% Ruffin 27% 19% Tondelier 39% 17% Villepin 29% 24% Wauquiez 35% 24% Zemmour 59% 13%

Each bar puts the two answers end to end, each with its own figure; the total length is not a published figure and is not shown. Candidates in alphabetical order.

All waves (very negative / rather negative)
Candidate29 Jul2 Sep
Gabriel Attal33% / 22%32% / 27%
Jordan Bardella41% / 12%41% / 11%
Xavier Bertrand30% / 22%29% / 23%
Bernard Cazeneuve26% / 23%27% / 23%
Olivier Faure31% / 21%28% / 24%
Raphaël Glucksmann29% / 21%28% / 21%
François Hollande35% / 26%37% / 27%
Sarah Knafo33% / 13%32% / 12%
Marine Le Pen43% / 12%42% / 10%
David Lisnard21% / 11%22% / 13%
Jean-Luc Mélenchon61% / 14%60% / 12%
Édouard Philippe32% / 22%31% / 26%
Bruno Retailleau31% / 20%32% / 19%
Fabien Roussel27% / 19%25% / 22%
François Ruffin28% / 20%27% / 19%
Marine Tondelier36% / 17%39% / 17%
Dominique de Villepin30% / 24%29% / 24%
Laurent Wauquiez36% / 22%35% / 24%
Éric Zemmour57% / 14%59% / 13%

How to read this chart. Each line follows one candidate: the higher it is, the larger the share of French people who reject them in the survey. For Elabe, each bar shows two shares side by side, very negative image then rather negative.

Learn more

The figures are those of the pollsters' reports, rounded to the point. The table under each chart gives the same values as text. Small gaps between waves stay within the sample's margin of error.

What this page does not say

Rejection is not a vote intention: someone can reject one figure without backing another.

The two series have neither the same question nor the same answers; their levels cannot be compared. Only the movement within one series, wave after wave, means anything.

This is not an estimate of the 2027 vote and P(win) does not rest on it.

How the model is tested on past elections · See also the full method