Robust analysis of bibliometric data

Abstract

The aim of the work is to reproduce the image of the research profile of the Italian statisticians derived from the querying of bibliometric databases. We highlighted the need for multiple sources in order to convey a truer picture and how data could be combined in order to have a classification or an index of the overall productivity, which took into account all sources and metrics. The data matrix contains a set of metrics from a variety of databases for each author and it is a sparse matrix (there are many zeros). Furthermore, the variables are leptokurtic and characterized by positive asymmetry. In order to apply the classical techniques of multivariate analysis, data must first be transformed or, alternatively, robust analysis techniques have to be used. In the paper we will focus our attention on this type of bibliometric data, describing their main characteristics and problems. In addition, a robust approach to the analysis of these data will be presented.

Co-authorship networks and scientific performance: an empirical analysis using the generalized extreme value distribution

“…Since individual h-index is a non-decreasing function of scientific career, a positive effect of age (as proxy of scientific age) should be expected. In our case, the non-significant effect of age is consistent with the results of De Battisti and Salini [7] on the publication style of the 444 Italian researchers in Statistics ('Statistics subfield' in our analysis). Analysing bibliometric data (number of publications, number of citations, and h-index), from several data sources (ISI-web, Scopus, Google Scholars, and CIS), they found that 'The productivity does not increase with increasing age.…”

A bibliometric-based evaluation on environmental research in the Arab world

“…It is also useful in making pronouncements related to qualitative features of scientific research output (Wallin 2005). This technique is valuable in deriving good information and knowledge about the status of scientific research activities in specific disciplines that help researchers to recognize novel schemes among research (De Battisti and Salini 2013). Furthermore, it is a well-recognized technique for conducting systematic analyses (van Raan 2005).…”

Bibliometric analysis on global Catha edulis (khat) research production during the period of 1952–2014

“…The evaluation of scientific research at global, regional, or national levels is necessary to improve its research productivity [ 18 ]. The bibliometric approach is a research method used to evaluate the current state of scientific research in certain areas by conducting a precise analysis to obtain statistics that can be considered indicators of achievement that allow researchers to recognise and improve their research [ 19 , 20 ]. This type of analysis utilizes quantitative methods and statistics to describe publications within a given field, journal, institution, or country [ 21 , 22 , 20 ].…”