Recently at the center of controversies for failing to provide accurate forecasts in several significant elections (the American presidential election, the Brexit referendum, the elections in Austria and the Netherlands) where the results defied predictions, pollsters achieved a small comeback last Sunday.
At last, it was the two predicted candidates who won, with scores fairly close to those they were credited with on the eve of the election.
In fact, polling institutes bear major political and economic stakes that prevent them from being considered simple, neutral sources.
Their surveys have become central to media discourse.
The number of surveys is experiencing an exponential increase today: it has doubled in twenty-five years and even increased tenfold in the case of electoral polls alone.
Heirs of the “flies” of the 17th century, which were deployed in public places to report overheard conversations, the methods have continued to become more systematic alongside the development of statistics.
In fact, despite the criticisms, the results of opinion polls retain a particularly strong, even growing, social influence, with significant political consequences.
Ultimately, we would be wrong to forget: at the foundation of every opinion poll lies a commercial contract between a service provider company and a client who will be billed.
And yes, unlike the thermometer and a fever, opinion polls have an influence on the phenomenon they claim to measure.*
