France’s problem with polls

John Lichfield is a former foreign editor of the Independent and was the newspaper’s Paris correspondent for 20 years.

CALVADOS, France — Opinion polls are influential in all democracies, criticized constantly but watched obsessively. In France, however, the polls are more than just influential; they have become part of the country’s electoral machinery. All the more reason, one would think, for French polls to be reliable. Increasingly, they are not. 

The relentless conveyor belt of polls, or sondages, released by the 14 competing organizations in France is confusing at the best of times. Before next year’s election? Even more so. 

So how popular is President Emmanuel Macron, eight months before he seeks to become the first French head of state to win a second term in two decades? According to different polling organizations, posing slightly different questions, Macron’s “popularity” or “confidence” or “approval” rating ranges from 36 percent to 45 percent. 

How popular is the far-right leader Marine Le Pen, who will probably be Macron’s strongest challenger next year? Every French polling organization foresaw a breakthrough for Le Pen in the first round of regional elections last month. Every poll got it wrong. 

Polls for the regional elections hugely inflated support for Le Pen’s renamed party, National Rally (RN), because the turnout was much smaller than expected. Surveys for round two also proved to be deeply flawed. 

This was not an isolated mistake. French polls have overestimated the strength of the far right in all seven rounds of the last four elections (presidential, European and local). French polls used to undercount the potential far-right vote. Now they systematically overcount it. 

The record of French polling is, on the whole, no worse than those in other countries. In big national elections, it has been reasonably reliable, apart from serious lapses in 1995 and 2002. 

Arguably, however, the sondages have a duty to be more accurate than elsewhere because they are even more powerful. They have increasingly become part of the apparatus of politics, rather than just a commentary on it. 

I don’t suggest — as some do — that the polling organizations are partisan. But I do believe that they have been absorbed into the electoral system in a way that is potentially dangerous. The combined effect of a dominant presidency, a weak parliament, two-rounds of voting and the decline of powerful, stable political parties mean that opinion polls can now shape events in France, rather than simply anticipate them. 

In 2017, opinion polls helped a young centrist upstart, Emmanuel Macron, reach the second round of the presidential election. They created the momentum that persuaded both center-left and center-right voters to migrate to Macron and away from their own, admittedly flawed, candidates. 

Similarly, in 1995, opinion polls showed that Lionel Jospin was best placed to fight the center right for the presidency. The polls, not his ideas or personality, were what won him the Socialist primary. The same happened with Ségolène Royal in 2006. 

Polls can have a similar magnetic pull and influence in U.S. presidential primaries, which are held within a rigid two-party system. In France, however, the power of the polls has been hugely magnified by the collapse of the old left-right duopoly and by the weakness of its political parties. 

In the coming months, French opinion polls will be more significant than ever, and they will play a decisive role in whether there will be a single center-right challenger to Macron and Le Pen. If such a candidate emerges, Macron could be knocked out in the first round. Otherwise — according to, well, the polls — he should have the relatively easy task of beating Le Pen in the runoff. 

Four or five center-right barons and one baroness (Valérie Pécresse, president of the regional council, Île-de-France) hope to be the sole candidate of the “traditional” right. Most of them want a primary election. One does not. 

Xavier Bertrand, president of the northern French region Hauts-de-France, says that he will not enter a center-right primary under any circumstances. And why should he? By refusing to participate, he has, in effect, created his own primary in which the result is determined by opinion poll ratings, where he is the front-runner. 

Polls now give Bertrand 16 to 18 percent of the first-round vote — not quite enough to reach the second round when Macron and Le Pen are at 22 to 28 percent. If, however, he begins to move up in the polls to 20 percent or more, he will emerge as the obvious and unavoidable candidate of the center right. 

That is his strategy. That is why he jumped into the race in early March. 

But how reliable are the polls? Four of the other would-be center-right contenders complained in an op-ed in Le Figaro last month that it was absurd to leave it up to opinion polls to decide on a candidate, especially when “the credibility of the polls is more than ever open to question.” 

They have a point. 

Comments by polling company officials point to two reasons why les sondages may have become less reliable in recent years — more so in local elections than national ones. 

First, there is the switch to the online polling of semi-permanent voter panels instead of random telephone calling. One polling company official told me: “Online polling is much, much cheaper. It offers the impression of testing a constant sample and their fluctuating opinions. But we may not be reaching the apolitical or disaffected voter in the way that random calling does.” 

Second, there is a growing disaffection with politics of all kinds — even within the anti-establishment far right. Turnout is falling, but it’s falling unevenly amongst different social and age groups. This is one reason why the far right vote has been systematically overestimated in the last four years. 

Marine Le Pen’s efforts to sanitize her father’s party may, paradoxically, be another explanation. Voters used to hide the fact that they voted for Jean-Marie Le Pen’s National Front. Now, they boast to pollsters that they intend to vote for Marine’s RN, but they eventually don’t — or at least they don’t bother to turn up. 

Opinion polls still have a place in a democracy. “They remain an essential tool for analyzing the situation,” says the doyen of French pollsters Jérôme Jaffré, head of the public opinion think tank Cecop. “Take them away, and you blind the media.” 

Others, like Philippe Guibert, political commentator and former head of the government information service, agree in principle but say that French polling has become too frantic and too dominant. “Political analysis has been reduced in many cases to a horse race based on the virtual and the hypothetical,” he said. 

Alain Garrigou, professor of politics as the University of Nanterre-Paris and author of L’Ivresse des sondages (The Drunkenness of the Polls), goes even further. He says that the power of the pollsters has undermined confidence in French politics and reinforced the impression that the system is controlled by and for an elite. 

That may be exaggerated. What is certain, however, is that a situation in which opinion polling is simultaneously more influential and less accurate is a dangerous prospect for French democracy.