| Description |
The literature suggests that good and competent politicians can improve government performance. This leads to the following question: what characteristics make specific political candidates preferred to others, increasing election chances? We create a new dataset – dubbed the "Open-Candidati-Europee"- to address this question – of all the 1076 candidacies for the European Parliament elections held in Italy in May 2019. The dataset assembles a rich set of characteristics regarding the candidates' profiles extracted from Curriculum Vitae and other administrative sources. The new dataset overcomes two limitations found in the literature: first, it also covers non-winning candidates; second, it accounts for the presence of candidates on the web and social media. The empirical analysis provides descriptive evidence that political experience and education positively predict candidates' success. We confirm previous findings of a gender bias towards female candidates. Furthermore, we find that the popularity on the web of the Candidate (Google trend, Facebook, Twitter) is positively associated with a candidate's success and can significantly reduce the effect of education, experience, and gender bias; we also describe how those effects vary within the political spectrum. Finally, we employ these characteristics to predict the electoral performance of candidates using state-of-the-art machine learning models.
|
ICAR Research Data Repository for Knowledge Management