Abstract
We exploit the recent introduction of a law for political transparency to create a new dataset of all the 1076 candidacies for the European Parliamentary elections held in Italy in 2019. The dataset – dubbed the “Open-Candidati-Europee” – assembles a rich set of characteristics regarding the candidates’ profiles as extracted from CV and other sources. This dataset provides a solution to overcome two important limitations for the study of political selection: we cover all the population of candidates (including non-winning candidates) and we include important omitted factors of candidates’ success (such as the presence on the web). To provide a first exploratory analysis, we leverage on the expected electoral results predicted by the position in the list to provide descriptive evidence that political experience, education, and presence on the web positively predict candidates’ success. We confirm previous findings of a gender bias towards female candidates. Finally, we show how to use these characteristics in a machine learning framework.
