Presidential 2022: did the polls influence the vote in the first round?

►”Polls focus attention on the game rather than the issues”

Alexandre Deze, lecturer in political science at the University of Montpellier, author of 10 lessons on political polls (De Boeck).

For the month of March 2022 alone, 121 surveys related to the presidential election were published. This is slightly less than in March 2017 (137), but it still represents an average of four polls per day, which is a lot. In fifty years, the number of polls for the French presidential election has multiplied by 40, to reach the figure of 560 during the last election in 2017. This shows how essential the media consider them to be.

→ SURVEY. Presidential 2022: in the factory of the polls

In the campaign, the polls were of major importance. First at the level of the selection of candidates: we have been able to observe, with the case of Éric Zemmour or Anne Hidalgo, the effects of qualification and disqualification that polls on candidates can produce. And this in particular because of the weakening of the parties which once exercised a monopoly in this area.

Second, it is likely that the surveys influenced voters’ votes. The campaign was particularly fluid, the debate inaudible, the agenda untraceable. In such a configuration, they probably played more of a compass role than usual. Ordinarily, nearly two-thirds of voters do not necessarily pay attention to the polls. But the debates aroused on the left around the useful vote for Mélenchon suggest that they have this time been taken into account by part of the electorate in the voting decision.

→ DEBATE. Should we do without electoral polls?

It should also be noted that, more than ever, political commentary has taken hold of the polls, transforming the presidential election into a horse race and focusing attention on the game rather than on the issues. The consequence is that the polls generated for a large part of the campaign balance of power quite far from the final result. Let us remember that last fall Éric Zemmour was given 19% of voting intentions and potentially present in the second round, while Marine Le Pen was announced as the future big loser of the election…

These errors mainly come from problems of representativeness of the samples and the consecutive operations of adjustment of the raw results. Under these conditions, one can wonder how the institutes have nevertheless managed to correctly give Emmanuel Macron, Marine Le Pen or even Yannick Jadot in their latest estimates. As such, I remind you that over the past twenty-five years, all electoral consultations combined, these estimates have proven to be approximate or erroneous in almost one out of two cases. This first round is therefore an average vintage for the pollsters.

►”The French system strengthens the influence of these decision-making tools”

Emmanuel Riviere, Director of International Studies and Policy Advisor at Kantar Public

Indeed, the polls influence the vote. More precisely, they are part of the elements that voters have at their disposal to choose from. However, the French electoral system, with the ax of the second round, reinforces the influence of these decision-making tools. Since in the second round there can only be two candidates, voters watch who is most likely to qualify. This is the notion of “useful vote”.

→ RESULTS. 27.84% for Macron, 23.15% for Le Pen… The final results of the 1st round

We can clearly see the effects of this in Sunday’s results, which show the crushing of all those who had no chance of reaching the second round and the postponement, from the first round, of their potential voters towards better placed candidates. in the polls. We thus observe a transfer of those who could have voted for Valérie Pécresse to Emmanuel Macron, and perhaps a little to Marine Le Pen, and a transfer of potential voters for Yannick Jadot, Fabien Roussel and Anne Hidalgo to Jean-Luc Mélenchon.

This may have also played, but more on the margins, for Emmanuel Macron: the very rapid progress of Marine Le Pen in the polls may have influenced part of the electorate, who may have decided to vote for his best-placed opponent. The notion of useful vote is particularly effective on the left where the voter, now free from any partisan ties, is often more available for several possible candidates.

→ EXPLANATION. Presidential 2022: how the polls weigh on an election

In short, what happened on Sunday is the logical consequence of what opinion polls were saying last week. In our last productions with our partner Epoka, we clearly identified who the hesitants were and what each candidate’s potential was, which clearly showed that nothing was set in stone. In general, the pollsters were much less mistaken than during the last regional, since they correctly described the dynamics of progression and regression of each other. We must be able to read the polls not as tools for predicting results but as tools for identifying possible scenarios.

Afterwards, that does not mean that there is nothing to improve. For example, are the panels representative of the French population? Some have said that the people to whom we send the questionnaires being voluntary, they are by nature more inclined to answer, and therefore less abstaining, less popular…

Among various avenues, we worked, at Kantar Public, on the establishment of a panel with random recruitment, Public Voice: we will look for people drawn at random, instead of it being the people who register. In addition, we ensure the political representativeness of our samples by verifying that their indications on what they voted for in 2017 do indeed make it possible to reconstruct the balance of power in the last presidential election.


Leave a Reply

Your email address will not be published. Required fields are marked *