The local and global structure of knowledge production in an emergent research field: An exponential random graph analysis

Gondal, Neha

doi:10.1016/j.socnet.2010.09.001

Cited by 38 publications

(22 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1 2 This sense of the "whole network" should not be confused with the technical term "complete graph," where every vertex has a direct edge to every other vertex. 3 [44] is still one of the best discussions of the interplay between the formal, statistical and substantive motivations for using exponential families.CONSISTENCY UNDER SAMPLING OF EXPONENTIAL GRAPHS 3 likelihood or pseudo-likelihood) to the observed sub-network, and then extrapolate the same model, with the same parameters, to the whole network; often this takes the form of interpreting the parameters as "provid [ing] information about the presence of structural effects observed in the network" [54], page 194, or the strength of different network-formation mechanisms; [2,16,17,24,25,55,62] are just a few of the more recent papers doing this. This obviously raises the question of the statistical (i.e., large sample) consistency of maximum likelihood estimation in this context.…”

mentioning

confidence: 99%

Consistency under sampling of exponential random graph models

Shalizi¹,

Rinaldo²

2013

Ann. Statist.

156

143

View full text Add to dashboard Cite

The growing availability of network data and of scientific interest in distributed systems has led to the rapid development of statistical models of network structure. Typically, however, these are models for the entire network, while the data consists only of a sampled sub-network. Parameters for the whole network, which is what is of interest, are estimated by applying the model to the sub-network. This assumes that the model is consistent under sampling, or, in terms of the theory of stochastic processes, that it defines a projective family. Focusing on the popular class of exponential random graph models (ERGMs), we show that this apparently trivial condition is in fact violated by many popular and scientifically appealing models, and that satisfying it drastically limits ERGM’s expressive power. These results are actually special cases of more general results about exponential families of dependent random variables, which we also prove. Using such results, we offer easily checked conditions for the consistency of maximum likelihood estimation in ERGMs, and discuss some possible constructive responses.

show abstract

mentioning

confidence: 99%

Consistency under sampling of exponential random graph models

Shalizi¹,

Rinaldo²

2013

Ann. Statist.

156

143

View full text Add to dashboard Cite

show abstract

“…Table ). Either those publications achieve higher citation rates in general or this is due to a regional effect, a “home‐field advantage.” For instance, Gondal () showed in a smaller and topic‐specific study that many authors are cited by their region and Cronin () conducted a study wherein American authors had a much higher tendency to cite American publications rather than British authors. We tested this assumption in Appendix and were unable to confirm this.…”

Section: Methodsmentioning

confidence: 99%

Placing articles in the large publisher nations: Is there a “free lunch” in terms of higher impact?

Schubert

Michels

2013

J Am Soc Inf Sci Tec

View full text Add to dashboard Cite

This paper deals with the role of a journal's publisher country in determining the expected citation rates of the articles published in it. We analyze whether a paper has a higher citation rate when it is published in one of the large publisher nations, the U.S., U.K., or the Netherlands, compared to a hypothetical situation when the same paper is published in journals of different origin. This would constitute a "free lunch," which could be explained by a Matthew effect visible on the countrylevel, similar to the well-documented Matthew effect on the author-level. We first use a simulation model that highlights increasing citation returns to quality as the central key condition on which such a Matthew effect may emerge. Then we use an international bibliometric panel data set of forty-nine countries for the years 2000-2010 and show that such a "free lunch" implied by this Matthew effect can be observed for top journals from the U.S. and depending on the specification also from the U.K. and the Netherlands, while there is no effect for lower-ranked American journals and negative effects for lower-ranked British journals as well as those coming from the Netherlands.

show abstract

“…Some of these networks are called two‐mode networks, where nodes represent different entity types. For example, Gondal () built a citation network consisting of paper and author nodes, and examined citation patterns between papers and authors. Keegan, Gergle, and Contractor () studied a Wikipedia network consisting of editor and article nodes.…”

Section: Related Workmentioning

confidence: 99%

“…Recent studies have used ERGM to study knowledge diffusion in various domains, such as citation networks (Gondal, 2011), multicultural research collaborations (Sayogo, Zhang, & Pardo, 2011), organizational information seeking (Johnson, Kovács, & Vicsek, 2012;Su & Contractor, 2011), and medical innovation (Zappa & Mariani, 2011). Table 2 summarizes these works based on the network dimensionality and parameter estimation methods.…”

Section: Exponential Random Graph Modelmentioning

confidence: 99%