Abstract

The introduction of big data technologies into political elections promises greater control of the electoral campaign at every stage: from defining the agenda, to manifacturing consent and monitoring results. Big data are, in fact, an effective tool for electoral profiling, especially when combined with structured databases and psychometric techniques. However, it is in the area of election forecasting that the real effectiveness of big data is being analysed, and the starting point is the search for the perfect algorithm to analyse heterogeneous data. This article intends to contribute to the debate via a critical analysis of the way big data is used to predict election outcomes. The focus of our analysis is the first round of the 2017 French presidential elections, and the voting forecasts produced during the campaign.