Considered the favourite for the European elections in the autumn polls, the far-right Rassemblement National party of Marine Le Pen is now predicted to finish in second place with 20% of voting intentions, behind Emmanuel Macron’s La République En Marche. EURACTIV France reports.
The RN now supposedly has a one-in-three chance of arriving first at the European elections, compared to a two-in-three chance for Emmanuel Macron’s party, if polls of polls are to be believed.
However, predicting the far-right party’s results is no easy task. Indeed, these have been systematically overestimated by the polls at the five most recent elections.
The difference between the pollsters’ predictions and the party’s election results is due to pollsters systematically overestimating the RN by around 2 percentage points. On the other hand, traditional parties’ results have been underestimated, even if only by a very small margin (between 0.2 and 0.5 percentage points), as shown by the graph below.
Scarred by the 2002 presidential election, when Le Pen’s father Jean-Marie finished a surprise second, pollsters have long corrected their assessment of far-right votes.
Voting for the far-right was considered such a taboo at the time that respondents did not dare admit their voting preferences to pollsters. But this is no longer the case: French people do not hide who they vote for anymore.
“We’re a very long way away from far-right being underestimated, which traumatised pollsters at the 2002 presidential election,” said Alexandre Andorra, who manages the website PollsPosition and has long reflected on the reasons for this overestimation by pollsters.
“In fact, the opposite is true: polls now systematically overestimate the far-right,” he added.
Abstention, a disease of European elections
So why do pollsters continue to systematically overestimate voting intentions for the RN?
According to Andorra, “this may be because pollsters know they’ll be criticised for underestimating the RN, but not for overestimating it, which may introduce a bias.”
Another explanation could be related to the rate of abstention. “Watch out, the European elections may hold some surprises, particularly from the smallest parties,” warned Aurélien Preud’homme, director of studies at Viavoice.
There is always a degree of uncertainty associated with abstention, especially for the European elections. Only 42% of French people voted at the last elections in 2014, and 2019 is not expected to be much different.
“Abstention is the most complicated factor to assess, it plays out over the last days,” Preud’homme said.
In case of high turnout, it is the established socialist, conservative and centrist parties that are expected to benefit. However, the RN itself has become established in the French political landscape and now enjoys a base of voters comprising about 20% of registered voters.
In contrast, other parties have seen their levels of support fluctuate considerably in recent years, with the emergence of LREM at the centre, the virtual disappearance of the French Socialist Party (PS) on the left and the collapse of the conservative Les Républicains party on the right.
“A high level of mobilisation by RN voters may therefore influence the outcome,” Preud’homme said.
According to Preud’homme, the overestimation of RN results by pollsters was related to standard margins of errors and the volatility of voting behaviour. Whereas French voters were once committed and loyal to one party, they now tend to dither and hover, making up their minds at the last minute.
This may be another reason to be less engaged with polls, which are too often seen as predictive tools.
Alexandre Andorra from PollsPosition called for a more considered approach to polls, which are often taken as gospel in France, with each new poll replacing the last one. In the space of two minutes, the whole of France can suddenly be convinced that Les Républicains will receive 12%, rather than 8%, at the European elections.
In reality, like all pollsters known, polls are just a photography of a sample of the population at a certain moment in time.
Moreover, these figures are only averages. “Statistically speaking, averages aren’t very meaningful: they can represent huge differences in probability,” Andorra cautioned.
[Edited by Zoran Radosavljevic and Frédéric Simon]