Anatomy of the generalized inverse Gaussian-poisson distribution with special applications to bibliometric studies

摘要

The vast number of observed bibliometric and scientometric datasets display a definite downward deviation from a straight line in the upper tail, when plotted in a double logarithmic coordinate grid. For this reason customary theoretical distribution laws are very poor representations of the observed phenomena. This disadvantage also extends to recently suggested models such as the Yule, the two- and the three-parameter Waring distributions. The main types of the GIGP distribution are described and two important limiting cases are discussed. The constrained minimum x2 method is developed for the estimation of the three parameters α, b, and γ. Finally it is argued that the Kolmogorov-Smirnov goodness-of-fit test is not applicable in the field of bibliometrics.