Archetypal Networks for Clustering Classroom Social Structures

Schools and classrooms are complex social systems in which individuals and contexts continuously interact. Peer relationships are one of the main components of this system that shape students’ behav-iors, beliefs and choices. Classifying classrooms based on their relational dynamics across different social layers can inform more effective edu-cational interventions. The present contribution proposes the use of Archetypal Analysis to cluster school-based network data. This method enables the synthesis of multiple classroom and school networks into a small number of “pure” and interpretable network configurations, high-lighting prototypical relational structures associated with various post-diploma choices, particularly university enrollment. Using primary data collected in a sample of high school students in Southern Italy,the approach clusters friendship and different kinds of support networks among classmates by considering structural and topological network indices, along with contextual variables, such as family background and school type.