Interfacing Network Simulations and Empirical Data
Abstract
This project tests Social Network Models for longitudinal data against empirical data using an original statistical test to determine the effectiveness of various models at reproducing networks. The Link Probability Model (LPM) is introduced as a viable model for the reproduction of social networks in dynamic equilibrium. We survey social network simulation packages and find that Construct uses a continually updated LPM as its stochastic engine, further establishing the LPM's viability as a social network model. We use actor oriented models to estimate statistically significant behavior on empirical networks and provide guidance on future extensions into multi-agent simulation, which is a rapidly growing area of research. Our findings rely on various empirical datasets and provide analytical results on the nature and structure of the social networks observed in them.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2009
- Accession Number
- ADA497468
Entities
People
- Anthony Johnson
- Ian Mcculloh
- Joshua Lospinoso
Organizations
- United States Military Academy