Standardizing relative impacts: Estimating the quality of research from citation counts
✍ Scribed by Van Hooydonk, G.
- Publisher
- John Wiley and Sons
- Year
- 1998
- Tongue
- English
- Weight
- 243 KB
- Volume
- 49
- Category
- Article
- ISSN
- 0002-8231
No coin nor oath required. For personal study only.
✦ Synopsis
The relative impact of local research units is obtained scores for local research units are deduced from relative by dividing the observed number of citations to their pubimpacts. These are obtained by dividing the observed lications by the expected number of citations. It is argued number of citations to their publications by the expected that the expected citation rates used in the standard number of citations to these publications 2 for a given method cannot lead to relevant bibliometric scores for period of time: The larger relative impact, the higher specific research topics. Extracting information about quality of research with the standard method is, there-quality of research. However, to make sense, citations and fore, almost impossible. The existence of empirical relapublications in nominator (local data) and denominator tions between the number of citations and the number (international data) have to be counted under the very of publications for scientific disciplines and for journals, same conditions. This condition for comparative citation leads to alternative ways to determine relative impact.
counts to be valid, is not fulfilled in the standard method, Hereby, reference data are taken from within a given research topic. Only observed citation and publication solely based upon publications, processed by ISI (the (activity) patterns for research topics are taken into ac-Institute of Scientific Information): The process of citing count for calculating bibliometric scores. The new methis subject-driven (applied for data in the nominator) ods are not restricted to ISI-publications. The resulting rather than journal-driven (used for the data in the debibliometric scores can contain information about the nominator). quality of research, and lead to different rankings than those obtained with the standard method, although the On the other hand, a short-cut bibliometric analysis same citation and publication data are used. based upon impact factors of journals, wherein research results are published, refers to a relatively short period of time (2 years). Given the average cited half-life of * This article is dedicated to Professor Greta Milis-Proost at the Theory occasion of her retirement in 1997. 1 Within the constraints of IRT (Item Response Theory), it was
The number of citations to publications will always recently shown that impact can be measured as a latent variable (Aldepend in some way on the number of publications and varez & Pulgarin, 1996) defined by a set of indicators (citations and publications) and leads to journal-rankings, different from those published by ISI (such as in ).
2 This is usually done by means of so-called expected citation rates, derived from all citations to all contributions of the same type, published in the same year in the same ISI-journal. By extension, discipline related