According to the most recent polls, Emmanuel Macron, leader of the En Marche! movement, will become the next president of France: He is forecast to win 61 percent of votes in the second, and final, round of the election on Sunday, against 39 percent for Marine Le Pen, of the far-right National Front. The die appears to be cast — at least according to numerous pollsters, and their reputation was recently burnished by the accuracy of their predictions for the first round on April 23.
But not so fast. Marine Le Pen could actually be elected president on Sunday with less — even much less — than 50 percent of voters’ “intentions de vote” (declared, intended votes) and without any extraordinary event occurring between now and then.
I’m a sociophysicist, and I’ve developed a model of how people’s opinions interact by applying to social behavior concepts and techniques from the field of physics. It allowed me to predict the Brexit vote in Britain and Donald J. Trump’s victory in the United States. Applied to the French presidential election, the model leads me to conclude, as have most researchers, that Ms. Le Pen cannot reach the 50 percent bar in voter intent.
But she doesn’t need to hit that threshold to become president. With, say, just 42 percent of voter intent in her favor, the outcome of the election remains very uncertain because of various forms of abstentionism, some extremely difficult for pollsters to anticipate.
For a better understanding of the sociological aspect of this phenomenon, let’s go back to April 23. From the beginning of the campaign, it was assumed that Ms. Le Pen would make it to the runoff. At stake in the first round, therefore, was the question of who would be the other finalist: Mr. Macron, François Fillon (Les Républicains) or Jean-Luc Mélenchon (La France insoumise)? Mr. Macron won, but his final score — 24.01 percent, compared with 21.30 percent for Ms. Le Pen and 20.01 percent for Mr. Fillon — wasn’t commanding enough to signal his ultimate victory. The rankings in that round could easily have turned out differently.
From these observations emerges an idea that is simple but deserves emphasizing: Voters’ opinions, far from being stable, were part of an ongoing dynamic, and a dynamic is not necessarily linear. It has multiple elements; it can speed up, slow down or even reverse course. The actual moment of voting therefore was decisive.
The same holds now, on the eve of the runoff this Sunday. Some of the people who were disappointed by the results of the first round — that is, supporters of Mr. Fillon or Mr. Mélenchon — now declare, with conscientious resignation, that they intend to vote for Mr. Macron. But that’s a choice fraught with ethical and political tension. In the name of rejecting Ms. Le Pen, people who have been denouncing economic liberalism or the record of the outgoing president, François Hollande, as causes of France’s problems — including the ascent of the National Front — will now supposedly vote for the man who embodies those tendencies. By voting for Mr. Macron, they would be endorsing a cause in order to ward off its effects.
At the same time, nearly two-thirds of about 240,000 Mélenchon supporters indicated in a mass consultation on May 2 that they preferred to abstain from voting or to cast a blank or void ballot rather than vote for Mr. Macron. Whether they actually hold to it or not, their stated position validates abstentionism, adding a third choice to a runoff normally limited to two options: the non-vote. (Polls predict an overall abstention rate of 25 percent, and that’s not counting blank or void ballots.)
As soon as some voters dare to flaunt their refusal to rally the so-called republican front, a concerted effort to defeat Ms. Le Pen, it becomes all the easier for others to somehow omit going to the polls on Sunday to vote for Mr. Macron or, once there, to cast a blank or void ballot. Call this an electoral Freudian slip, or a kind of unavowed abstention.
The phenomenon is almost the inverse of the one that played out in the early days of the National Front. Back then, voting for the far right was hard to own up to; it was shameful. This is partly the reason that pollsters were taken by surprise when Jean-Marie Le Pen, the party’s leader at the time, qualified for the runoff in the 2002 presidential election. After several election cycles, they have learned to account for that hidden variable and weight it properly. And today, National Front supporters are for the most part committed, and dedicated to the party’s program — they are reliable voters.
Instead it’s abstentionism that is now difficult to measure, especially unavowed abstentionism. Whereas a shameful vote cast for the National Front back in the day was a premeditated act, unavowed abstention against the National Front today would be a more or less accidental act of omission. Election day on May 7 will be the first time this choice can be expressed; unavowed abstention is, by nature, undetectable by polls.
I’ve devised a little mathematical formula to assess its potential effect on Sunday. Let’s assume, based on the latest polls, rates of “intentions de vote” of 39 percent for Ms. Le Pen and 61 percent for Mr. Macron, and rates of effective participation — that is, the proportion of voters who really will vote as they say — of 90 percent for Ms. Le Pen and 60 percent for Mr. Macron. In that case, Mr. Macron will win. But if effective participation for Ms. Le Pen hits 95 percent, whereas Mr. Macron’s doesn’t reach 61 percent, then he’ll lose. The differential in the rates of unavowed abstention between the two candidates could allow Ms. Le Pen to become president with far less than a majority of intended votes.
Also note that, according to my calculations, every point gained in intended votes for Ms. Le Pen has a multiplied effect on the participation threshold that Mr. Macon would have to meet in order to be elected. To take the last example: An increase of three points in intended votes for Ms. Le Pen, from 39 percent to 42 percent, would require the effective participation rate for Mr. Macron to jump by 8 points, to 69 percent, in order for him to win.
I am not claiming any of this will happen on Sunday. I am just saying that it could. And I am saying that the polls, while not actually mistaken, do not adequately reflect the weight of sociological behavior that is measurable in other ways, such as unavowed abstentionism.
Serge Galam is a physicist, a researcher at the Centre national de la recherche scientifique (CNRS) and a member of Cevipof, a political science research institute at Sciences Po. This essay was translated by John Cullen from the French.