Marginal models with individual-specific effects for the analysis of longitudinal bipartite networks

Area of Science:

• Social Network Analysis
• Statistical Modeling
• Network Science

Background:

• Bipartite social networks are complex systems.
• Analyzing relational events over time presents challenges.
• Existing models may not fully capture actor behavior dynamics.

Purpose of the Study:

• To propose a novel modeling framework for bipartite social networks.
• To analyze sequences of partially time-ordered relational events.
• To interpret actor participation and cooperation tendencies.

Main Methods:

• Directly modeling the joint distribution of actor involvement in events.
• Using a parametrization based on first- and second-order effects.
• Employing composite likelihood inference and classification for actor clustering.

Main Results:

• The model effectively represents latent individual behavior trajectories.
• Second-order effects reveal tendencies for cooperation.
• First-order effects indicate individual participation tendencies.
• The approach is validated on simulated and real-world data.

Conclusions:

• The proposed framework offers a robust method for analyzing dynamic bipartite social networks.
• It provides interpretable insights into actor behavior and network structure.
• Applicable to various fields, including scientific collaboration analysis.