Once again, the “official” polling institutes failed to anticipate François Fillon and cast a shadow over an entire profession!
Once again, we were right!
In our Friday article, we warned you that Alain Juppé, “touted” as assured of being in the second round, could have lost part of his electorate to the benefit of François Fillon, considered a more acceptable or less objectionable candidate than Nicolas Sarkozy.
We were the only ones to envisage this scenario, just like in 2002 when we cautioned voters about the possibility of JM Le Pen being in the second round of the presidential election.
That was 14 years ago and nothing has changed!!
What would it take to avoid polls that are so disconnected from the result?
First of all, a good dose of humility, by not making polls say what they don’t and better training political commentators in the analysis of political polls.
Moreover, it’s time to ask the right methodological questions:
Is an internet poll, especially in the context of a presidential election where we know seniors are the most likely to vote, a good tool?
Is a sample of fewer than 1000 people, presenting a low confidence interval with a multitude of candidates (7 for the Republican primary), sufficient?
Questions asked that absolutely do not take into account the fact that more and more voters are no longer voting for a candidate but for the candidate who will block another?
These re-evaluations have never been made despite repeated failures. Why?
Simply because an internet poll costs almost nothing to produce, and adding depth to a questionnaire in an attempt to understand voter motivations also costs more to produce.
Unfortunately, it seems to us that this is the price to pay to regularly avoid providing future voters with erroneous data that influences their vote, biases the democratic process, and discredits all polling institutes and the media that disseminate them.
by Frédéric Ganneval (ArteNice Consulting)
