Macrostructure from Microstructure: Generating Whole Systems from Ego Networks.

Area of Science:

• Network Science
• Computational Social Science
• Statistical Modeling

Background:

• Global network inference from sampled data is challenging.
• Existing methods may not accurately reconstruct complex network structures.
• Ego network sampling offers a practical approach to data collection.

Purpose of the Study:

• To develop and validate a new simulation method for global network inference.
• To reconstruct unknown network features using sampled ego network data.
• To compare the performance of the proposed method against baseline models.

Main Methods:

• Utilizing Exponential Random Graph Models (ERGM) for network reconstruction.
• Employing sampled ego network data as input.
• Conducting validity checks on established network datasets (Add Health, Sociology Coauthorship).

Main Results:

• The proposed method successfully reproduces key network properties like distance and main component size.
• The simulation method demonstrates superior accuracy compared to simpler baseline models.
• The study validates the efficacy of ERGM in global network inference from sampled data.

Conclusions:

• The novel simulation method provides a robust approach to global network inference.
• Ego network sampling combined with ERGM is a viable strategy for understanding large-scale networks.
• Further research should explore the limitations and extensions of ego network sampling.