Mixed Effects Dynamic Network Actor Models
Alvaro Uzaheta, Viviana Amati, Johannes Wachs, Christoph StadtfeldDynamic Network Actor Models (DyNAMs) assume that an observed sequence of relational events is the outcome of an actor-oriented decision process defined by the time until an actor initiates the next relational event and the choice of the receiver of the event.
The choice of the receiver is modeled through the maximization of the utility function of a specific receiver choice. In current DyNAMs, the utility function is constant over actors, unless their differences can be directly measured. However, this assumption can be unrealistic in situations where unobserved sources of heterogeneity exist.
We introduce the Mixed Effects DyNAM (ME-DyNAM), an extension of DyNAMs that allows modeling actor heterogeneity by including random-effects parameters. We illustrate the applicability of the ME-DyNAM by analyzing sequences of "who follows whom" in a community for aspiring and professional graphic designers. These relationships have been argued to matter as they are a primary source of information and inspiration on creative platforms of that kind. We study how behavioral patterns of follower relationships vary between platform users.