Wikipedia provides an excellent summary graphic (above) of opinion polling in the 2017 French Presidential election (first round). It is illegal to publish any more opinion polls after midnight last night, so this is the final picture we have of the state of the race prior to the election. Of course with any polling, one has to issue a number of caveats:
Firstly, the opinion polls called both the Brexit and Trump elections wrong. Yes Hillary Clinton did win the popular vote by c. 3%, but polls had her winning by a higher margin, and predictions were based on state level polling and these called the crucial swing states of Pennsylvania, Wisconsin, Michigan and Florida wrong. Given these predictions were based on hundreds of polls, this was an industry level failure.
Similarly, Brexit polls overwhelmingly showed the Remain side to have a narrow lead although some polls did show a majority for leave, and the trend line showed a significant convergence in the last few weeks and days of the campaign:
In both cases pollsters faced the problem of how to predict the behaviour of late breaking undecided voters in the race who appear to have plumped heavily for Brexit and Trump. There may also have been a shy Tory effect, whereby respondents may have been reluctant to share their real preferences with pollsters, for fear of being seen to make an unfashionable or unpopular choice.
Pollsters in the French Presidential elections may have similar difficulties, with even recent polls showing a level of undecided or abstention in the 20-30% range. How these potential voters behave will be crucial to the eventual outcome, given that the four leading contenders are all within a narrow 5% range.
There may also be large scale tactical voting with voters voting not necessarily for their preferred candidate, but for a candidate they perceive to be at least minimally acceptable, in order to ensure they will have at least one acceptable candidate to vote for in the second round.
This could damage candidates such as Hamon who are seen as being unlikely to make the second round, with Parti Socialiste voters opting for Mélenchon or Macron instead. Similarly some Fillon supporters may opt for Macron if they feel the latter has a better chance of making the second round and if they do not wish to be faced with a choice of Le Pen or Mélenchon at that stage.
Finally, there is the impact of late breaking events such as the terrorist incident in the Champs Elysees last Thursday. Will this help more right wing “Law and Order” or racist candidates like Le Pen or Fillon? Only three opinion pollsters were in the field on Friday, and appear to have found no discernable impact on voter intentions. But in fairness to pollsters, it is very difficult to discern how this might impact the outcome on Sunday.
But if we were to take the polling trends at face value, what outcome can we expect? The graphical trendline of polls (see top image) appears to show Macron establishing a small lead over Le Pen at the top of the leader board, with Fillon maintaining an even smaller lead over a surging Mélenchon who appears to be benefiting from Hamon’s rapid decline and elimination as a likely second round contender.
Le Pen has been declining steadily in the polls perhaps because Mélenchon is also eating into her anti-EU support whilst also becoming a credible second round candidate. Despite some last minute jitters in the Macron camp, Macron appears to be holding his overall lead in the contest.
So the most likely second round scenario would therefore appear to indicate a Macron/Le Pen contest, with Macron winning easily by a 62-37% margin. Macron would also beat Fillon by a 65-35% margin or Mélenchon by a 60-40% margin. Mélenchon would also beat Le Pen by c. 60-40% or Fillon by 56-44%, and Fillon would beat Le Pen by c. 57-43%.
It thus appears that the French people would prefer almost anyone to Le Pen in the second round, with Macron slightly favoured to be her opponent and ultimate victor in the second round. However Macron’s support is also viewed as being the “softest”, with many people unclear as to his policies and viewing him as, at best, the lesser of four evils. He doesn’t have a long established party machine to support his candidacy and is therefore most at risk of under-performing his polls.
With 20-30% of voters still undecided or likely to abstain, any combination of the leading four candidates is still possible. We are thus likely to see a President Macron, Fillon, Mélenchon, or Le Pen in declining order of probability.
How French Polling has faired in the past:
I agree with you generally: Marcon looks like he has broken into a lead.
If I took the polling at face value the main question is whether Fillon or Melenchon can get past le Pen, but in historic terms the gap between Le Pen and the other two is on the higher end of the historic polling error in France.
Still anything could happen.
Agreed. The largest error recorded in the sample period is 3.36%, which would not be enough to lift Fillon or Mélenchon above Macron or Le Pen. So it would be a bit of an upset if either made the second round at this stage. But as fjallstrom says in the European Tribune version of this diary, the problems with telephone polling may have been getting gradually worse…
Good catch by fjallstrom. I have wondered myself about phone polling. Landlines are increasingly a thing of the past – I have mine only because I get a bundle discount for doing so. Otherwise, it’s all mobile phones, and how many of us are willing to answer from an unfamiliar or blocked number? The samples we get left with may not reflect reality as well as they once did.
Market polling has adepted many years ago. See Ipsos. The failure of political polls is for other quite practical reasons, voters who don’t remember who they voted for, no party affiliation and vast number who decide in the voting booth. Getting the demographics right, the turnout and what happens with populist elections or referenda, a large percentage of people who go in for the very first time to vote.
A number of Dutch political polls have done so for more than a decade. Polling by Maurice de Hond makes corrections for underrepresented demographics which adds another layer of fault margin.
○ Opinion Polls Score Much-Needed Dutch Election Win
Might describe the behavior of political junkies (something on the order of 1% of voters), but have yet to see any evidence of this among the other 99% of voters in actual elections. The most “strategic” not fully committed other voters seem to get is calculating whether or not to vote at all.
This isn’t accurate. Macron has MoDem and the conservative wing (including Hollande), of PS behind him.
Unlike the UK pure choice between Leave (nationalist/euro-skeptic) or Remain (globalist) the choices in France (excluding UPR struggling to get 1%) are mixed bags.
Don’t forget France like many European countries has a proportional voting system where tactical voting or voting in order of your preference is not uncommon.
Not operative in France’s presidential election. Although there is a public financing component with a divide above and below 5%. So, would guess that is a consideration for PS voters.
Yes but there is a culture of tactical voting that may be less prevalent in polities which use only first past the post systems.
Yeah, that’s why half of PS voters switched to a non-existent new party to consolidate with MoDem in the presidential election.
Looks like the polls weren’t too bad. The candidates polling as the top two indeed were the top two – Macron (who had held a narrow lead in polls) and LePen. Now we get the real test: Macron is supposed to have a comfortable lead in the second round. Will be curious as to how polling trends over the next couple weeks and how accurate the pollsters are once all is said and done. Beyond that, as always, hoping that Macron receives healthy support come election day. We need as many firewalls against right-wing nationalism as possible.
Interesting tidbit from Nate Silver:
I think Brexit and Trump have served as a salutary reminder to all voters that indulging in fantasies, not bothering to vote, or playing “lefter than thou” games can have serious consequences, and that we can’t take long cherished norms or institutions like the EU for granted.
Interesting table embedded in tweet linked to above. European far right parties have often underperformed their opinion poll averages. More recently, Brexit and Trump may also have served as a salutary reminder to European voters that indulging in fantasies, not bothering to vote, or playing “lefter than thou” games can have serious consequences, and that we can’t take long cherished norms or institutions like the EU for granted.
Candidate Vote Poll average (based on graphic at top of diary)
Macron 23.9 24.0
Le Pen 21.7 22.3
Fillon 20.0 19.7
Mélenchon 19.2 18.9
All in all, a very good performance by the pollsters with Fillon and Mélenchon slightly out-performing their polls and Le Pen underperforming by the largest margin.
The second round may be more challenging to predict. With a 25% advantage in the opinion polls at present, Macron may have difficulty motivating his supporters to vote again, and left wing voters could abstain en masse. Or will Le Pen voters become demotivated by the size of Macron’s lead?
It will be interesting to see whether he maintains that lead in the polls in the run up to the second round.
From my diary – France: Macron Wins – National Vote Estimate [Update] .
○ ‘En Marche!’ – Youth and Hope for France