+ 3 points for Valérie Pécresse at Harris Interactive, + 11 points according to Elabe, after his victory at the LR congress. A Marine Le Pen who, a little less than a month ago, ahead of the Republican candidate for Ifop-Fiducial, while for Opinion Way, it was just behind. Enough to give a big headache to the French who no longer know where to turn in this slump of figures regularly published by polling institutes, in view of the presidential election.
“The criticism of the polls is as old as the poll itself, believes Thomas Vitiello. On the other hand, confidence in their results has indeed been very relative over the past twenty years. Doctor in political science and teacher at Sciences Po, the researcher, formerly of the Elabe polling institute, notes a rise in distrust of the polls, which he attributes to an increased distrust of the media and political power.
“The institutes have no interest in manipulating the results”
However, the production of electoral polls is governed by very strict rules, and their results supervised by the Survey Commission, responsible for ensuring the accuracy of the figures listed. “The institutes have no interest in manipulating the results they publish, assures Thomas Vitiello, because, for them, the electoral polls are their showcase. If they represent only between 5% and 15% of their turnover, the surveys published during the electoral campaigns are indeed the most visible.
The reliability of the polls cannot therefore be called into question. The difference in results from one survey to another, all carried out according to the quota method – which consists of bringing together a sample of people that is the most representative of the French population – would be mainly due to calculation variables from one institute to another. “There are different ways of measuring the same thing,” explains the researcher. The published results will therefore not necessarily be calculated on the same basis. »
Short-lived results but lasting trends
The population samples of the institutes on which voting intentions are assessed are most often between 1,300 and 1,400 people only. With response categories that are not necessarily the same from one institute to another. Differences in assessment which would therefore explain the sometimes significant disparities between the results of two polls published a few days apart.
The ” tips by Thomas Vitiello to find your way around? “Comparing results from one institute to another is misleading. To be able to identify trends, it is better to compare polls from the same institute,” he says. If a survey is only valid at the moment T, the comparison over a long time allows above all to have an overview of the dynamics of each candidate. Since the start of the campaign, the broad lines of voting intentions in the first round of the presidential election have remained more or less the same with Emmanuel Macron still in the lead, a second place being played out between Marine Le Pen and Valérie Pécresse, and a left increasingly divided at the end of the procession.