Driven by data; ridden with liberty.
Emmanuel Macron has been elected as the President of France.
(Video: AFP News Agency)
Similar to the first round, I will look at the final polls from each company, that were conducted on the 4th and 5th May 2017.
These polls were conducted by: Ipsos (5th May 2017), Harris (4th – 5th May 2017), Ifop-Fudicial (2nd – 5th May 2017), Odoxa (4th May 2017), Elabe (4th May 2017) and OpinionWay (28th April – 4th May 2017).
There were also two polls including 4th May in their fieldwork dates, but were not the final polls from those houses, and so were excluded from the calculation: Ipsos (4th May 2017) and Ifop-Fudicial (1st – 4th May 2017).
Weighing the polls by sample size gives a final poll average vote share of 62.5% for Macron. (A straight average of the final six polls gives 62.3%.)
This is an error of 3.6% (or 3.8%, if preferring the unweighted average), which is surprising, given the astounding precision of the first round polling.
Importantly, there was no poll between the first and second rounds ever gave the centrist Macron 66%. This may be a function of post-polling swing, since the second round election was held on the 7th.
Despite the error being roughly the same in size as for the EU referendum, and larger than the 2015 British general election, it is unlikely this vote share error will induce an intensive inquiry into the state of French polling.
No-one cares when you underestimate a landslide.