Inferring global network properties from egocentric data with applications to epidemics

Area of Science:

• Network Science
• Sociology
• Epidemiology

Background:

• Social networks are complex systems often incompletely observed.
• Inferring global network characteristics from local, egocentric data is a significant challenge.

Purpose of the Study:

• To investigate the relationship between egocentric network data and global network properties.
• To determine the extent to which global network properties can be inferred from partial observations.

Main Methods:

• Analysis of different types of egocentric data.
• Characterization of global network properties, specifically the largest connected component (giant) and epidemic outbreak size.
• Asymptotic analysis assuming uniform network selection.

Main Results:

• Egocentric data generally provide a wide range of possible sizes for the giant and epidemic outbreaks.
• These findings indicate that egocentric data carry minimal information about these specific global network properties.
• The asymptotic sizes of the giant and outbreak are mathematically characterized.

Conclusions:

• Egocentric data are insufficient for accurately estimating global network properties such as the giant component and epidemic spread.
• Network scientists and epidemiologists should be cautious when inferring large-scale network behavior from localized data.
• Further research may be needed to develop methods that can extract more information from egocentric network samples.