The revised SNIP indicator of Elsevier's Scopus

TL;DR

This paper discusses the revised SNIP indicator by Elsevier, highlighting its improvements over previous versions and addressing issues of transparency and statistical robustness in scientometric evaluations.

Contribution

It analyzes the modifications in the SNIP indicator, emphasizing the challenges of using arithmetic averages with skewed data and the implications for evaluation transparency.

Findings

The revised SNIP addresses some problems of the original.

Using averages in skewed distributions is problematic.

The indicator's opacity affects evaluation transparency.

Abstract

The modified SNIP indicator of Elsevier, as recently explained by Waltman et al. (2013) in this journal, solves some of the problems which Leydesdorff & Opthof (2010 and 2011) indicated in relation to the original SNIP indicator (Moed, 2010 and 2011). The use of an arithmetic average, however, remains unfortunate in the case of scientometric distributions because these can be extremely skewed (Seglen, 1992 and 1997). The new indicator cannot (or hardly) be reproduced independently when used for evaluation purposes, and remains in this sense opaque from the perspective of evaluated units and scholars.