EXIT POLLS. According to initial estimates, Emmanuel Macron would again win his confrontation with Marine Le Pen in the second round of the presidential election. With a smaller gap than five years ago, however.
Presidential results near you
[Mis à jour le 24 avril 2022 à 23h57] End of suspense. Emmanuel Macron won the presidential election tonight and will therefore return with a second term. According to estimates made by Ipsos Steria for France 2, he would have won 58.8% of the vote, against his competitor Marine Le Pen who would peak at 41.2%. However, these percentages will be refined over the night until the total count of the votes. The Linternaute teams inform you live of the evolution of theare presidential results.
There is therefore no longer any need to refer to the results published in foreign media, in particular the Swiss media Le Temps, the Tribune de Genève, the RTS, but also the Belgian media Le Soir, La Libre Belgique or the RTBF which have in the past been customary the publication of estimates before 8 p.m., legal time in France. In fact, they are not subject to the same constraints as the French media, which must respect a “reserve period”. Freed from this law, these Swiss and Belgian newspapers have devoted themselves to covering the French ballot during election days for about ten years and have published precise estimates of the election result. During the last presidential election, for example, they had given the names of the two finalists with a percentage of votes close to the final result.
Is the ban on the dissemination of “exit polls” respected?
The eight main polling organizations, namely BVA, Elabe, Harris Interactive, Ifop, Ipsos, Kantar, Odoxa and OpinionWay had agreed not to carry out exit polls during the first round of the election which took place on April 10. It was a request from the polls commission, which had precisely indicated in a press release that it had obtained from these institutes the assurance that none of them would carry out “exit polls” on April 10. The commission was very clear, if not intransigent, towards a leak of results before legal time: “It follows that any reference, on election day, to such polls can only be the result of rumors or manipulation and therefore no credit should be given to them”.
And yet, the risk of leaking trends on the results of the second round of the presidential election remains high: in 2017, when the polls commission urged French polling institutes not to communicate anything about their sampling work, media Swiss and Belgians had given the results of the first round around 4 p.m. And this time again, during the first round of 2022, figures circulated from the end of the afternoon on foreign sites, in particular Belgian ones, so that La Libre Belgique published at 5:39 p.m.: “According to a first survey polls, Emmanuel Macron and Marine Le Pen would be tied for the lead. They would both collect 24% of the votes, learned La Libre from a good source, “before specifying that these were only” results from a first exit poll, carried out by one of the major institutes”. Which would mean that one of the organizations did not respect this (more or less tacit) agreement concluded with the polls commission. However, no evidence has been verified.
Why are exit polls so rare?
Among the three methods used by the polling institutes, namely the exit poll, the telephone poll and estimates, only the third is still used. Exit polls are no longer fashionable because of the biases they generate. In fact, their reliability is easily questionable: if a pollster questions you, you can answer something other than your real vote, for fear of judgment from others in particular. A respondent who is surrounded by acquaintances, whether relatives or neighbours, may hide his real opinions because he does not want to reveal them in broad daylight. To protect himself, he will interchange his actual vote with the one he feels is most acceptable and accepted by the people on his side at the time. This problem is thus linked to the declarative nature of this type of survey, and it can end up completely biasing the response when the panels are reduced. For this reason, more and more polling institutes are reluctant to use this method and prefer a count of “real” ballots during the first counting, in several hundred “test” polling stations. While the polling stations are opening for the second round, it should be noted that the “exit polls” polls that you will see circulating today are not carried out by the main French institutes. So it’s up to you to take your precautions.
Jean-François Doridot, managing director of Ipsos Public Affairs France, also assured the newspaper Le Monde that “there are no more real exit polls, for financial reasons. This type of operation presents a exorbitant cost for a qualitatively lower result compared to the estimates produced by all the polling institutes for 20 hours”. But he specifies that other institutes carry out polls on the day of the vote relating rather to the sociology of the voters. “These polls give a result during the day and can sometimes leak, but, on the one hand, they are not made public by the institutes and, on the other hand, if they are polls, they must be read with the same precautions as the polls published a few days earlier,” he adds.
How are exit polls conducted?
How can the polling institutes give results from 8 p.m. when the counting has not started in the major cities of France? One of the solutions is the establishment of what are called “exit polls”. The principle is quite simple: an investigator from the institute questions a large number of voters who have just voted, by going to several polling stations in a municipality, considered “representative” of the population registered on the lists. This method developed from the 1980s and was quickly modernized, allowing the most serious institutes to propose, from 8 p.m., figures fairly close to the final figures provided several hours later, after counting the votes, by the Ministry of Interior.
The polling institutes establish their first estimates not from polls, but from the counts carried out in test polling stations, closing at 6 p.m. Few agree to deliver the manufacturing secrets of these estimates. But some agree to give the number of polling stations used to establish them. At Ifop, the results of “nearly 300 offices” were for example analyzed and extrapolated during the last presidential election to give a national trend, according to its director Frédéric Dabi. At OpinionWay, between 200 and 300 test polling stations were scrutinized. At Ipsos, we went “from 200 to 250 polling stations to 500” for the previous presidential election, to properly trace the information. The goal: to have enough test offices to have fallback solutions if some of them do not report their results quickly enough. Finally, at Kantar, the number of test polling stations is part of the “manufacturing secret”.
Do the figures released at 8 p.m. by the media have a link with the “exit polls” polls?
At 8 p.m., the media broadcast the figures collected in collaboration with polling institutes: for example, France Télévision-le Parisien with Ipsos, BFMTV-l’Express with Elabe, TFI with Ifop, les Echos-Radio Classique with OpinonWay , or Harris Interactive with M6-RT. These are estimates based on the first counts of the polling stations closing at 7 p.m.: generally, the big cities are not concerned. In the first round of the election on Sunday April 10, these elections were not available until 7:30 p.m. or 7:40 p.m. They were therefore not polls, and even less “exit from the polls”, -as explained by the Check News of Liberation on April 12 -, since the circulation of these figures on foreign sites or social networks such as Twitter kicked off in the afternoon. They were released earlier and therefore cannot have originated from these estimates. It will be the same for this second round: the results revealed at 8 p.m. by the media will not come from ‘exit polls’.