Uncertainties and Ambiguities in Percentiles and how to Avoid Them

Introduction The recently proposed fractional scoring method (Schreiber, 2012) has been discounted by Leydesdorff (2012) as computationally intensive and as unnecessary complex. While his arguments are correct, I do not find his solution of attributing papers to percentile rank (PR) classes convincing. A Leydesdorff admits, his approach leads to an uncertainty depending of the applied counting rule. Although this uncertainty decreases with the number N of documents as 1/N, I find this uncertainty unsatisfactory.