A dynamic model for the mutual constitution of individuals and events. (3rd March 2022)

• Record Type: Journal Article
• Authors: Lerner, Jürgen; Lomi, Alessandro
• Editors: Perc, Matjaz
• Is Part Of: Journal of complex networks. Volume 10:Number 2(2022)
• Publication Date: 2022-03-03
• Subjects: duality of persons and groups -- two-mode networks -- hypergraphs -- longitudinal network models -- relational hyperevent models -- Deep South; Numerical analysis -- Periodicals; Computer networks -- Periodicals; Social networks -- Periodicals; 518.05
• DOI: 10.1093/comnet/cnac004
• Languages: English
• ISSNs: 2051-1310

Abstract

Abstract: We argue and show that a recently derived class of relational hyperevent models (RHEM) may be adopted to extend the sociological concept of duality by linking it to empirical data containing information on the temporal order of events. We show how RHEMs may be specified to predict the likelihood that combinations of individuals of any size will jointly participate in future events, conditional on their history of participation in past events. We show, further, how RHEMs may support hypothesis testing about competing mechanisms driving participation in events. Finally, we show how RHEMs may be used to establish the location of the events that actually happened in the much larger space of all the possible events that could have happened, but did not. We illustrate the empirical value of RHEMs using a canonical dataset containing information on the participation of 18 women in 14 time-ordered events. We provide dynamic network visualizations to link empirical estimates of the model parameters to qualitative insight on the dynamics of the mutual constitution of individuals and events. While RHEMs are also applicable to large networks (e.g. coauthorship networks), using a small canonical dataset allows us to examine in greater detail the model's implications for each and every observed event and to identify the location of each event participant in the network of previous events. Scaling down our model to examine a small dataset affords a more detailed understanding of the link between quantitative model results expressed as parameter estimates, and the qualitative features of the original observations.