An evaluation of exponential random graph modeling and its use in library and information science studies

Publication date: Available online 6 September 2016 Source:Library & Information Science Research Author(s): Ana Dubnjakovic Social network analytical tools and theories have long been an accepted part of the research landscape in many social and physical sciences including: sociology, political science, psychology, communications, business, geography, biology, physics, and chemistry as well as library and information science (LIS). Given the level of activity in the social network analysis (SNA) area concerning LIS, it is important to review the latest trends in the SNA stochastic modeling, namely exponential random graph (ERG) models. Unlike previous SNA methods, ERG models offer insight into generative network properties through simultaneous inclusion of structural parameters and attributes in the analysis while accounting for the interdependent nature of network data. Additionally, when Monte Carlo Markov Chain Maximum Likelihood Estimator is used, ERG modeling results in parameter estimates superior to other methods (e.g., MRQAP). The current study will demonstrate the utility of ERG models in LIS through a brief overview of major concepts and techniques in SNA, followed by a detailed description of ERG modeling technique, a review of currently available software used in analysis and a brief examination of its current use in LIS studies.
Source: Library and Information Science Research - Category: Databases & Libraries Source Type: research