Testing social network metrics for measuring electoral success in the italian municipal campaign of 2011

It is often argued that the bias hidden in Social Media data prevent from using them for any statistical inference. In this paper, we investigate the practicability of a new method for predicting electoral outcomes that is less affected by demographics and self-selection bias. In particular, we put in place a first test to understand which social network analysis metrics can exhibit positive correlation with electoral success. Our analysis is not intended to use social media audience as a sample of the whole electorate but just as a sample of the supporters of a candidate. In conclusion, we speculate on the information we can extract measuring the social network of the groups of supporters. Essentially, we get an overview on the variety and extent of the segments of the population represented in these groups, and this probably correlates with the capacity to attract consensus.