The poisson-lognormal model for bibliometric/scientometric distributions

摘要

The Poisson-lognormal model assumes that the intensity parameter of a Poisson process has a lognormal distribution in a sample of observations. This model can yield highly skewed, discrete distributions, but must be estimated by numerical methods. When applied to many of the empirical data sets related to the ‘laws’ of Lotka, Bradford, and Zipf, this compound Poisson model produces good to excellent fits. Discussion includes possible ‘causal’ processes and some implications for future bibliometric and scientometric studies.