Cahier 2016-22

Title:The modelling of networks using Exponential Random Graph Models: an introduction
Abstract:Networks are representations of relational data. Whether the data used represents social interactions, cooperations or inter-bank dependencies, the structure of the network reflect a decision-making process based on many factors (common friends, technological proximity, geographical proximity). One of the objectives of network analysis is to identify these factors. The analysis of the structure using Exponential Random Graph Models (ERGM) can help in the identification of these factors by answering why two particular agents interact with one another, or why a specific agent has a particular position inside a network. In other words ERGMs allow us to perform an econometric analysis on network data. In the case of networks it is possible that a link depends upon the structure of the network. Usual econometric methods cannot be used because the dependence violates the hypothesis of independence of observations. ERGMs take into account this particularity of network data. The aim of this paper is to present the statistical theory behind ERGM models and present an application using R-Project.
Keyword(s):Network analysis ; ERGM ; Network structure ; R
Auteur(s) :Johannes VAN DER POL
JEL Class.:C59;C13;D85

TÚlÚcharger le cahier

Retour Ó la liste des Cahier du GRETHA (2016)

© GREThA 1999-2018