The use of the generalized Waring process in modelling informetric data

Abstract

Although its use in informetrics dates back at least to 1987, data analysed in a recent paper by Shan et al. (2004) has rekindled interest in the generalized Waring distribution (GWD). The purpose of this note is to show that for many purposes, the distribution is best motivated via a familiar informetric scenario of a population of “sources” producing “items” over time leading to a stochastic process from which the univariate, bivariate and multivariate forms of the GWD are natural consequences. Earlier work and possible future applications are highlighted. Many of the results are due to Irwin and Xekalaki while much of the material on the Waring process has been previously available in an unpublished research report by the author (Burrell, 1991).