Community detection in the European Parliament: A network approach

In this dissertation, we study networks of legislators, based on their voting behaviour. We describe different network-construction methods and we review the state of the art of how scholars transform voting data into networks describing the voting similarity of legislators. We describe how one can detect dense sets of nodes called “communities” and we discuss previous analyses of voting behaviour in the European Parliament. In addition to our review material, we include original research in a novel analysis of polarization. We use modularity optimization to measure polarization and detect communities in the 7th Assembly (2009–2014) of the European Parliament (EP). Our results suggest that polarization levels in the EP are different for different policy areas. We find relatively “robust” partitions that split the EP in two communities. It depends on the policy area whether we can assign a label to each community. For some policy areas (“environment”, “gender equality”), we find “left wing” and “right wing” communities, but for other policy areas (“agriculture”, “industry”), there appear instead to be an “extreme” community and a “centrist” community. Hence, our results suggest that one should take into account the differences between policy areas when one wants to study community structure in the European Parliament.