MAKING THE UNSEEN VISIBLE: EXPLORING CROSSCUTTING SOCIAL MEDIA PUBLICS AND THEIR SOCIOPOLITICAL TRAITS
Keywords:Facebook, digital trace data, sociopolitics, clustering, classification
This paper proposes an approach for studying the sociopolitical traits of multiple publics on Facebook that emerge in the network of interactions between users and public pages. The study is based on a survey of 1697 Danish citizens whose responses are coupled with their public Facebook activity.This is used to make predictions about a selection of sociopolitical features for a random sample of 50.000 Facebook users across more than 20.000 public pages. The interactions of the 50.000 users are modeled as a network and a clustering algorithm is used to find groups that arise naturally within said network. This allows for the study of how certain sociopolitical features cut across different congregations of the public in a way that retains a lot of the complexity of the digital trace data.
Results show that voting intention overlaps most strongly with the clusters in the network, followed by gender and geo-location. Additionally they show that the so-called political echo-chambers consist only of smaller subsections of the entire network with many users' interactions mainly being identified by interests that can be attributed to gender, geo-location or other. Although, results also show that the political alt. right are very dominant on hot button political issues such as immigration and religion.
It is proposed that by eliciting sociopolitical trends while considering the full network of interactions might lead researchers to overlook and overestimate fewer features when studying the formation of social media publics.