Multilevel longitudinal analysis of social networks.

Area of Science:

• Social network analysis
• Statistical modeling
• Developmental psychology

Background:

• Stochastic Actor-Oriented Models (SAOMs) are established for network dynamics analysis.
• Existing models often struggle with multilevel network panel data.
• Understanding interdependence in dynamic social networks is crucial.

Purpose of the Study:

• To extend SAOMs for analyzing multilevel network panel data.
• To develop a Bayesian approach for estimating complex network dynamics.
• To investigate the interdependence between friendship networks and minor delinquency.

Main Methods:

• A random coefficient model was integrated into the SAOM framework.
• Bayesian estimation techniques were employed for model fitting.
• The model was applied to longitudinal data from secondary school classrooms.

Main Results:

• The extended SAOM successfully analyzed multilevel network panel data.
• The model demonstrated the dynamic interdependence between friendship formation and minor delinquency.
• The Bayesian approach provided robust parameter estimates for network processes.

Conclusions:

• The proposed multilevel SAOM is a powerful tool for studying complex social network dynamics.
• This framework facilitates testing theories of social influence and interdependence across multiple networks.
• The findings offer insights into the interplay of social relationships and behavior in adolescent networks.