We study data from the unique EuroVotePlus (EVP) experiment with the help of a facial recognition software package to assess whether the facial features of European candidates can explain their popularity among voters. Within the weeks preceding the 2014 European election, we asked thousands of European web users to cast a simulated online vote for pan-European lists of candidates. The lists were built by randomly picking actual European deputies belonging to one of the seven political groups of the incumbent parliament. Each subject was presented with a ballot containing seven lists of ten candidates and then asked to successively vote under three different voting rules: closed-list, open-list with preferential votes, and open-lists with panachage. We use facial recognition software (FaceReader) to assess (i) characteristics of the candidates’ official picture and whether (ii) these characteristics explain their success in our experiment.
