Specifying Social Weight Matrices of Researcher Networks : The Case of Academic Economists

Abstract

This paper shows how collaboration and citation networks can be used to specify social weight matrices for a community of researchers. I use two competing theories of social influence on individual behavior, namely communication and comparison. I argue that in research networks collaboration captures communication, while citation captures comparison. I further argue that the comparison principle is likely to be the main social driver of individual research productivity and suggest a benchmark social weight matrix based on this principle. I test the benchmark matrix against several alternatives using a Bayesian model comparison procedure and conclude that the benchmark matrix outperforms alternative specifications. This result lends support to socio-economic theories underling the benchmark specification of the social weight matrix.