Exponential random graph modeling of a faculty hiring network: The IEOR case

Faculty hiring networks consist of academic departments in a particular field (vertices) and directed edges from the departments that award Ph.D. degrees to students to the institutions that hires them as faculty. Study of these networks has been used in the past to find a hierarchy, or ranking, amo...

Full description

Saved in:
Bibliographic Details
Published in:IIE transactions 2020-01, Vol.52 (1), p.43-60
Main Authors: del Castillo, Enrique, Meyers, Adam, Chen, Peng
Format: Article
Language:English
Subjects:
academic analytics
Apexes
Clustering
Graph theory
hierarchical networks
latent location graph model
Networks
Operations research
Ranking
Social networks
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Faculty hiring networks consist of academic departments in a particular field (vertices) and directed edges from the departments that award Ph.D. degrees to students to the institutions that hires them as faculty. Study of these networks has been used in the past to find a hierarchy, or ranking, among departments, but they can also help reveal sociological aspects of a profession that have consequences in the dissemination of educational innovations and knowledge. In this article, we propose to use a new latent variable Exponential Random Graph Model (ERGM) to study faculty hiring networks. The model uses hierarchy information only as an input to the ERGM, where the hierarchy is obtained by modification of the Minimum Violation Ranking (MVR) method recently suggested in the literature. In contrast to single indices of ranking that can only capture partial features of a complex network, we demonstrate how our latent variable ERGM model provides a clustering of departments that does not necessarily align with the hierarchy as given by the MVR rankings, permits to simplify the network for ease of interpretation, and allows us to reproduce its main characteristics including its otherwise difficult to model presence of directed self-edges, common in faculty hiring networks. Throughout the paper, we illustrate our methods with application to the Industrial/Systems/Operations Research (IEOR) faculty hiring network, not studied before. The IEOR network is contrasted with those previously studied for other related disciplines, such as Computer Science and Business.
ISSN:2472-5854
2472-5862
DOI:10.1080/24725854.2018.1557354