Identifying Friendship Layers in Peer Interactions on Campus From Spatial Co-Occurrences

• In Event: Advanced Statistical Methods for Complex Data

Abstract

This paper explores the network structure of peer interactions on campus among 3,910 freshmen enrolled in five large introductory courses. Peer interactions were inferred when two users were connected to the same WiFi access point during the same time window. Inspired by the Dunbar’s numbers, we applied clustering analysis on colocation time to identify friendship layers and the number of friends in each layer. Our results showed that most students had 5 to 6 layers of friends. The size of their closest circle ranged from 2 to 5 peers. This study offers new ways of thinking about students’ social networks that go beyond a binary view of social ties in educational literature.

Authors

• Quan Nguyen, University of Michigan - Ann Arbor

• Christopher Brooks, University of Michigan

• Oleksandra Poquet, University of South Australia