Authors:
Zsófia Boda | ETH Zurich | Switzerland
Timon Elmer
András Vörös
Christoph Stadtfeld
An extensive line of research into the evolution of friendship networks in communities emphasizes the importance of endogenous structural processes, individual characteristics, and meeting opportunities. First, much has been discovered about how endogenous structural processes such as reciprocity or transitivity shape friendship networks, inducing dependencies between network ties. Second, the role of individual characteristics is well-studied: people prefer others similar to them in their sociodemographic, behavioral, and intrapersonal attributes. Third, meeting opportunities for dyads or groups are essential in stimulating friendship formation between individuals.
While structural mechanisms and individual characteristics can be considered more or less stable, at least in an initial period, the quantity and quality of meeting opportunities seems more arbitrary. This is crucial, since even little changes in initial network patterns could lead to significant differences over the evolution of the network. Investigating the randomness in meeting opportunities and its effect on social ties is thus a key to better understand social network dynamics.
Using a unique combination of survey, observational, and experimental methods, our aim is to better explain the long-term role of initial random factors in friendships and dislike relations. For this, we focus on the complete network of over 200 first-year students at a Swiss university starting their studies together at the same department. We investigate network dynamics of the first semester (3 months). Survey data were collected in two different ways. First, detailed surveys, including questions about social ties, individual social background, and personal and behavioral characteristics, were administered three times: during the students’ first week; four weeks later; and during the last week of the semester. Second, mini-questionnaires were sent out 21 times (twice a week), collecting information about students’ interactions with each other. Observational data were collected using RFID-tags, which recorded actual face-to-face interactions during a socializing weekend at the beginning of the semester attended by almost third of the students. An experimental element has also been added to the study design. Three months before the semester started, freshly admitted students had a chance to attend an informal informational event at the university. Part of the event was organized in small groups, for which we randomly sorted students into groups.
Data were analyzed using longitudinal network models. Results show that the initial grouping of students still has an effect on social ties several months months later. This provides evidence that randomness in initial meeting opportunities can indeed strongly influence network evolution.