MODELING SOCIAL NETWORKS FROM SAMPLED DATA.

Area of Science:

• Social network analysis
• Statistical modeling
• Network science

Background:

• Network models are essential for understanding relationships in systems.
• Traditional social network analysis often assumes complete and error-free data.
• Real-world network data collection, especially through surveys, frequently involves sampling and potential measurement errors.

Purpose of the Study:

• To develop theoretical and computational frameworks for network inference using sampled data.
• To address the challenges posed by incomplete network observations and measurement errors.
• To adapt inference methods for social network models to sampled network information.

Main Methods:

• Review of practical network sampling designs.
• Development of a typology for network data within a likelihood inference framework.
• Formulation of inference methods for social network models using adaptive network designs.
• Analysis of link-tracing sampling designs' impact on collaboration networks.

Main Results:

• Established a conceptual and computational theory for inference from sampled network data.
• Proposed a typology of network data for consistent treatment in likelihood inference.
• Developed inference methods applicable to adaptive network designs.
• Demonstrated the utility of the methods through analysis of a collaboration network.

Conclusions:

• Inference from sampled network data is feasible and necessary for realistic social network analysis.
• Adaptive network designs and link-tracing sampling can be effectively incorporated into network inference.
• The developed framework provides a robust approach to handling incomplete and imperfect network data.